Loading video...
Video Failed to Load
Introducing EXPO-FT – Efficient, Reliable & Open-Source VLA Finetuning! EXPO-FT unlocks π0.5 for challenging manipulation tasks: Routing string lights & inserting the power connector to illuminate them Striking pool ball into pocket Inserting flower into wine bottle (1/5)
79,711 views • 4 months ago •via X (Twitter)
17 Comments
Unlike prior works that only train lightweight policies, rely on latent space prediction, or decoupled RL training, EXPO-FT fully finetunes the VLA with RL — leveraging EXPO's stability and sample efficiency, augmented by human-in-the-loop corrections (2/5)
Using an average of 19.1 minutes of online robot data, EXPO-FT reaches 30/30 on these tasks: Routing/powering string lights Striking a pool ball into a pocket Inserting flower into a wine bottle Scooping candy Picking up cube from large initial states Flipping egg (3/5)
EXPO-FT outperforms highly performant prior RL approaches by fully leveraging the VLA for online learning (4/5)
Project with @khhung906, @TianGao_19, @DorsaSadigh, @chelseabfinn Website: Paper: Open-source code coming soon! (5/5)
Really cool work :)
Cool work! I'm looking into exactly this atm. What are tasks you tried the method on where it didn't work well enough and that you didn't include in the paper?
Cool idea congrats!!!
Incredible work!
This is what I was looking for. Awesome work!
I don’t get it, this isn’t really finetuning a vla right? You’re finetuning an edit policy that updates the vla output? Eitherway good work but was just confused, what’s the improvement on prior work? Or is the only change is the fact that the policy is a vla
The VLA becomes the base policy in EXPO and is fine-tuned
awsome
Impressive results – 30/30 on 8 tasks with only 19 min of RL data. Quick question: these results are tested on visual variations within the same lab setup. How does EXPO-FT perform when the environment itself changes – different kitchen layout, unfamiliar objects, or a completely new cultural context? Is that a data problem or a method problem?
amazing
The most interesting AI progress isn't happening on screens anymore. It's happening in the physical world. Getting a robot to reliably plug in a connector or insert a flower into a bottle is often harder than generating a thousand lines of code.
can we say this is "fully finetuning" the VLA with RL? seems that you freeze original VLA during RL process and just modify the edit residuals actor and Q-function
Next you need to try with a thread and a needle😅 Looking forward to the open source too
![Heartfelt Limited-Time Resonance: Resonate 10 Times, Get 10 Free | Infinity Nikki Version 2.9 Outfit Preview: Florabound Promise It began with a flower awakening in the forest. As the wind rose, blossoms, skirts, and morning light swayed as one. A fairy's legend drifted through the mist, while radiant wishes took root in the shadows, until all the world was drawn into a dream in full bloom. >> Main Attribute: Sweet >> Outfit Ability: Sylvan Glow After using the ability, Nikki channels the life force flowing through the forest and enters the Forest Radiance state. This state remains active until logout or until the ability is manually disabled. After equipping Sylvan Glow, Nikki can consume energy to draw nearby creatures toward her, continuously attracting insects in the surrounding area. Once enabled, the attraction effect remains active until logout or until the ability is manually disabled. Attracted insects manifest as forest butterflies that follow Nikki. Each butterfly can carry 5 insects, and up to 6 butterflies can exist at once. Another outfit ability can then be used to catch those following insects. In Co-op, you cannot see or catch other stylists' forest butterflies. While in the Forest Radiance state, its energy can be used to power regional abilities in Itzaland. When using the outfit abilities Default: Gigantification, Dawn Stroll Sticky Claw, Default: Machine Control, or Skystride Dragon Statue, Forest Radiance energy is automatically consumed once the original energy is depleted, allowing the effect to continue. When in the Forest Radiance state, energy regenerates automatically over time and can also be replenished by collecting Blings. ❖ The ability Sylvan Glow cannot be used in certain stages or in the Mira Crown Merit Arena. >> Decoration: Moondew Blossoms Can be placed in open spaces. Moonlight is woven into the sweetest dreams, and deep within the woods, garlands bloom to the sound of song. If your radiant soul remains untarnished, fear not the figure dancing among the trees, for her gifts ask nothing in return. ❖ Available after reaching 160 Resonances in the Heartfelt Limited-Time Resonance Where Flowers Wake. >> Heartfelt Limited-Time Resonance: Where Flowers Wake During the event, select [×10 Resonate] in "Where Flowers Wake" to immediately trigger one bonus 10-draw Resonance. You can receive up to 100 free Resonances in total! Bonus 10-draw Resonances also count toward both guarantee and cumulative draw counts. ❖ During this Heartfelt Limited-Time Resonance, the bonus 10-draw can only be triggered by selecting [×10 Resonate]. Resonances made through [×1 Resonate] do not count toward triggering the bonus. Please take note. ❖ This bonus can be triggered up to 10 times. Once the limit is reached, further 10-draw Resonances will no longer trigger a bonus. >> Evolution Rewards ◆ Evolving the outfit to Stage 1 unlocks the pose Dawn Butterfly Whisper. ◆ Evolving the outfit to Stage 2 unlocks Sunlit Sway for Framed Moment. You can also use Stamen Heartchime in [Clothing Evolution > Decoration: Radiance] to obtain the decoration Moondew Blossoms: Radiance. ◆ Evolving the outfit to Stage 3 unlocks Sunlit Vow: Makeup, a new idle animation, poses, enhanced special effects, and alternative details for certain pieces. ♾Follow Infinity Nikki and quote this post with your UID, and we'll select 3 stylists to win $20 each. ❖ Winner announcement: We will announce the winners via the Infinity Nikki account by 23:59:59 (UTC-7) on September 7, 2026, and notify them via private message. —— ➤ Infinity Nikki Version 2.9 Bloom Beneath Bright Skies launches globally on August 27 (UTC-7)! ➤ Download Now: ➤ Join us on Discord: #InfinityNikki #BloomBeneathBrightSkies](https://image.24vids.com/tw-2091737333452415074/amplify_video_thumb/2091731362319663104/img/--nBtEtnBySvtnul.jpg)
![THE SEVEN SEALS The same day the Church is taken out of this world at the Rapture, judgement begins. As it was in the days of Noah, the same day Noah entered the ark, the judgement began. The same day Lot left Sodom, fire came from heaven and destroyed them all. So it is, on the same day that the Church leaves this world, at the Rapture, the Judgment begins. SEAL 1 - The Conqueror (The Revealing of the Antichrist). The Man on the White Horse. Rev 6:2 And I saw, and behold, a white horse, and he that sat thereon had a bow; and there was given unto him a crown: and he came forth conquering, and to conquer After the rapture of the Church,2 Thess.6v7 the first thing that happens in the earth is that the antichrist is revealed; a white horse and its rider. This is not the Lord Jesus- i.) “He that sat on the horse had a bow”, Jesus doesn’t use a bow, when you see Jesus, He uses a Sword. ii.) “A crown was given to him”- Jesus is King of Kings; He gives crowns. Rev.6v2 iii.) “He went forth conquering and to conquer”. The Lord Jesus is only going to battle against the Nations at the second coming, which is not yet So this is not Jesus, this is the antichrist. He had a bow, without arrows. He would conquer nations without firing a shot, he would conquer nations with peace; they would give their nations to him. Daniel 8:23-24 describes a powerful king who will arise at the end of a period of time, characterized by transgression and wickedness. This king will have a fierce appearance, understand dark schemes, and wield mighty power, not by his own means, but through extraordinary destruction. He will succeed in what he does, destroying the mighty and the holy people. SEAL 2- Conflict on Earth: Killings, Murders (The Man on the Red Horse) Daniel 6v3- 4 When He opened the second seal, I heard the second living creature saying, “Come [b]and see.” 4 Another horse, fiery red, went out. And it was granted to the one who sat on it to take peace from the earth, and that people should kill one another; and there was given to him a great sword. Once the Church is gone at the Rapture, no more peace on the earth, there would be killings. Matt.24v7,10 For nation will rise against nation, and kingdom against kingdom. And there will be famines, [a]pestilences, and earthquakes in various places. 8 All these are the beginning of sorrows. 9 “Then they will deliver you up to tribulation and kill you, and you will be hated by all nations for My name’s sake. 10 And then many will be offended, will betray one another, and will hate one another. Rev.13v15 15 He was granted power to give breath to the image of the beast, that the image of the beast should both speak and cause as many as would not worship the image of the beast to be killed. SEAL 3- Scarcity (Inflation and Famine) The Man on the Black Horse Daniel 6v5-6 When He opened the third seal, I heard the third living creature say, “Come and see.” So I looked, and behold, a black horse, and he who sat on it had a pair of scales[c] in his hand. 6 And I heard a voice in the midst of the four living creatures saying, “A [d]quart of wheat for a [e]denarius, and three quarts of barley for a denarius; and do not harm the oil and the wine. Lamentations 4v4-10 AMP [4] The tongue of the infant clings To the roof of its mouth because of thirst; The little ones ask for food, But no one gives it to them. [5] Those who feasted on delicacies Are perishing in the streets; Those reared in purple [as nobles] Embrace ash heaps. [6] For the [punishment of the] wickedness of the daughter of my people [Jerusalem] Is greater than the [punishment for the] sin of Sodom, Which was overthrown in a moment, And no hands were turned toward her [to offer help]. [7] Her princes were purer than snow, They were whiter than milk [in appearance]; They were more ruddy in body than rubies, Their polishing was like lapis lazuli (sapphire). [8] Their appearance is [now] blacker than soot [because of the prolonged famine]; They are not recognized in the streets; Their skin clings to their bones; It is withered, and it has become [dry] like wood. [9] Those killed with the sword Are more fortunate than those killed with hunger; For the hungry pine and ebb away, For the lack of the fruits of the field. [10] The hands of compassionate women Boiled their own children; They became food for them Because of the destruction of the daughter of my people [Judah]. Ezekiel 4v10-12, 17 AMP [10] The food you eat each day shall be [measured] by weight, twenty shekels, to be eaten daily at a set time. [11] You shall drink water by measure also, the sixth part of a hin; you shall drink daily at a set time. [12] You shall eat your food as barley cakes, having baked it in their sight over human dung.” [17] because bread and water will be scarce; and they will look at one another in dismay and waste away [in punishment] for their wickedness. If you think there are economic problems right now in the world, this is nothing compared to what is coming. No one would be able to solve the economic problem then, not even the antichrist. He would make promises, but would fail. This is going to be a very terrible time on earth. At this point the peace treaty is broken and the next three and a half years begins SEAL 4- Widespread Death on Earth (Death Released) A Pale(green) Horse with a rider called- Death, and Hell Followed. Rev.6v7 -8 7 When He opened the fourth seal, I heard the voice of the fourth living creature saying, “Come and see.” 8 So I looked, and behold, a pale horse. And the name of him who sat on it was Death, and Hades followed with him. And [f]power was given to them over a fourth of the earth, to kill with sword, with hunger, with death, and by the beasts of the earth. These are the judgments of God, not even the actions taken by the antichrist. The world is being punished for their rejection of Jesus Christ. Killings with the sword, refers to war. Matt 24v21 For then shall be great tribulation, such as was not since the beginning of the world to this time, no, nor ever shall be.” (KJV) The Fourth and Fifth Seals are in the second 3 and a half years, after the Peace treaty has been broken. 1 Thess. 5v3 -4 For when they shall say, peace and safety; then sudden destruction cometh upon them, as travail upon a woman with child; and they shall not escape. At this time the antichrist enters into the temple in Jerusalem and from then on, fierce persecution begins. SEAL 5- The Cry of the Tribulation Saints (Killed for their Faith) Rev 6v9-11 When He opened the fifth seal, I saw under the altar the souls of those who had been slain for the word of God and for the testimony which they held. 10 And they cried with a loud voice, saying, “How long, O Lord, holy and true, until You judge and avenge our blood on those who dwell on the earth?” 11 Then a white robe was given to each of them; and it was said to them that they should rest a little while longer, until both the number of their fellow servants and their brethren, who would be killed as they were, was completed. When a godly man dies, he goes to heaven and his soul is without a body, but is clothed with a spiritual garment until the resurrection. At the rapture, those who were dead in Christ were raised up with a glorified body, and we that are alive and remain were caught and changed into that same glorified body. The tribulation saints who were slain for their faith, they gave their lives as a sacrifice to God, they refused to take the mark of the beast, and were killed. Their souls are seen in Heaven under the Altar, without their spiritual glorified bodies, hence their cry. Vr 11 the Lord told them to rest for a while until others to be killed like them also arrive. At this time too something happens, there is more preaching of the gospel ongoing in the earth, Evangelism intensifies. Rev.14: 14 And I looked, and behold a white cloud, and upon the cloud one sat like unto the Son of man, having on his head a golden crown, and in his hand a sharp sickle. 15 And another angel came out of the temple, crying with a loud voice to him that sat on the cloud, Thrust in thy sickle, and reap: for the time is come for thee to reap; for the harvest of the earth is ripe. Here the Tribulation Saints are Raptured, which is getting to the ending part of the tribulation period. #YLWSPECIALSSeason1Phase3 #RaptureModeActivated #RaptureReady #FirstFlightGang #YearOfCompleteness](https://image.24vids.com/tw-1917164818424881604/amplify_video_thumb/1917140299744206848/img/9nAYVGEHiBzuy13A.jpg)
![i'm looking for feedback, thank you here's the full definition of my schematic: import { mkdirSync, writeFileSync } from "node:fs" import path from "node:path" import { defineCircuit, definePart, exportKiCadNetlist, instantiate, net, pin, } from "../../../tools/circuitd/src" const ch32v203f8u6 = definePart({ id: "CH32V203F8U6", kicadSymbol: "tinybee:CH32V203F8U6", footprint: "tinybee:QFN-20_L3.0-W3.0-P0.40-BL-EP1.7", defaultValue: "CH32V203F8U6", datasheet: " description: "144 MHz RISC-V MCU with ADCs, op-amps, advanced timers, and SWD", fields: { "LCSC Part": "C7570477", }, designNotes: [ "Power this MCU from a 3.3 V rail only; do not allow the VDD pin to see more than 3.6 V.", "Place a 100 nF decoupler immediately next to VDD and the QFN ground return, and keep that loop as short as possible.", "Put a small series resistor between PA0 and any off-board throttle or one-wire configuration signal so the MCU pin is not directly exposed at the connector.", "Keep PA13 and PA14 easy to probe and free of hard loads so SWD still works during bring-up and recovery.", "Keep PA8, PA9, PA10, PA7, PB0, and PB1 on the three phase-drive nets, and keep PA6/BKI free for a future fault or protection input.", "If you follow the openwch RISC-V ESC zero-cross scheme, tie PA3 and PA4 back into PA2 externally and reserve PA2 for that shared interrupt net.", "Route back-EMF and op-amp related pins away from phase copper, gate-drive loops, and other fast-switching nodes.", ], pins: { "PA0/WKUP/ADC0": "1", "PA1/ADC1": "2", "PA2/ADC2/OP2O0": "3", "PA3/ADC3/OP1O0": "4", "PA4/ADC4/OP2O1": "5", "PA5/ADC5/OP2N1": "6", "PA7/ADC7/OP2P1/CH1N": "7", "PB0/ADC8/OP1P1/CH2N": "8", "PB1/ADC9/OP1O1/CH3N": "9", "PB10/OP2N0": "10", "PB11/OP1N0": "11", "PB14/OP2P0": "12", "PB15/OP1P0": "13", "PA8/CH1": "14", "PA9/CH2": "15", "PA13/SWD/PA12/UDP": "16", "PA14/SWC/PA11/UDM": "17", "PA10/CH3": "18", VDD: "19", "PA6/ADC6/OP1N1/BKI": "20", GND: "21", }, }) const tlv75533pdqnt = definePart({ id: "TLV75533PDQNT", kicadSymbol: "Regulator_Linear:TLV75533PDBV", footprint: "tinybee:Texas_X2SON-4_1x1mm_P0.65mm", defaultValue: "TLV75533PDQNT", datasheet: " description: "500mA low-dropout fixed 3.3V regulator in 1x1mm X2SON-4", designNotes: [ "Treat +BATT as a 1S Li-ion or LiPo rail only; this regulator has a 5.5 V maximum input rating, so 2S or higher is out of bounds for this graph.", "Place at least 1 uF ceramic directly at IN and at least 1 uF ceramic directly at OUT; do not push those capacitors away from the package.", "Check regulator heating with (VIN - 3.3 V) * load current and add copper area if the dissipation is not comfortably safe.", "Add local input bulk when the battery lead is long or inductive so the LDO input does not absorb line spikes by itself.", "Keep EN tied to a known state at all times; if startup control matters later, break it out deliberately instead of bodging it in.", "The X2SON thermal pad is internally tied to GND; flood it into the local ground copper and do not leave the center pad floating.", ], pins: { OUT: "1", GND: "2", EN: "3", IN: "4", THERMAL_PAD: "5", }, }) const controlHeader = definePart({ id: "CTRL_IN_HEADER_1X01", kicadSymbol: "Connector_Generic:Conn_01x01", footprint: "tinybee:CTRL_PAD_1x01_Micro", defaultValue: "CTRL_IN", description: "1-pin control input wire pad for throttle signal", designNotes: [ "Use this only for the control signal.", "Ground reference is expected to be shared elsewhere in the system when this pad is in use.", "Battery power comes in through the dedicated battery wire holes, not through this control pad.", "PA0 on the CH32 is not a true FT input; treat this header as a 3.3 V logic input unless you deliberately add a real level-conditioning stage.", ], pins: { PIN1: "1", }, }) const motorHeader = definePart({ id: "MOTOR_OUT_HEADER_1X03", kicadSymbol: "Connector_Generic:Conn_01x03", footprint: "tinybee:MOTOR_PADS_1x03_Micro", defaultValue: "MOTOR_OUT", description: "3-pin motor phase wire pad group", designNotes: [ "This is the board edge interface for the three motor phases.", "Use direct wire holes here rather than a bulky 2.54 mm header footprint.", ], pins: { PIN1: "1", PIN2: "2", PIN3: "3", }, }) const programmingHeader = definePart({ id: "PROG_HEADER_1X04", kicadSymbol: "Connector_Generic:Conn_01x04", footprint: "tinybee:PROG_PADS_1x04_Micro", defaultValue: "PROG", description: "4-pin fine-pitch SWD programming header", designNotes: [ "Break out SWDIO, SWCLK, GND, and 3.3 V so the CH32 can be flashed and recovered without bodge wires.", "Keep this header free of extra loading and avoid reusing the SWD pins elsewhere until firmware bring-up is stable.", "Treat the 3.3 V pin here as target reference unless you deliberately design reverse-current-safe back-powering through the regulator path.", ], pins: { PIN1: "1", PIN2: "2", PIN3: "3", PIN4: "4", }, }) const batteryHeader = definePart({ id: "BATT_IN_HEADER_1X02", kicadSymbol: "Connector_Generic:Conn_01x02", footprint: "tinybee:BATT_PADS_1x02_Micro", defaultValue: "BATT_IN", description: "2-pin battery wire pad pair", designNotes: [ "Use this dedicated pair for battery positive and battery negative input.", "Keep the battery loop tight to the bulk capacitors and half-bridge supply entry.", ], pins: { PIN1: "1", PIN2: "2", }, }) const complementaryHalfBridge = definePart({ id: "PMCPB5530X_115", kicadSymbol: "tinybee:PMCPB5530X,115", footprint: "tinybee:DFN2020-6_L2.0-W2.0-P0.65-BL", defaultValue: "PMCPB5530X,115", datasheet: " description: "20 V complementary N/P MOSFET half-bridge in DFN2020-6", fields: { "LCSC Part": "C552747", }, designNotes: [ "Use this complementary half-bridge only on a 1S Li-ion or LiPo rail; do not reuse the direct P-gate pull-down topology above the 1S battery range.", "With +BATT limited to the 1S range, the direct P-gate pull-down swing stays inside the device gate limits and gives usable drive headroom.", "Drive the low-side N-FET gates directly from the 3.3 V timer outputs only in this 1S design; rework the stage before raising the battery voltage.", "Do not skimp on local +BATT bypass; keep real ceramic bulk plus a high-frequency bypass capacitor tight to the half-bridge supply loop.", "Pour the duplicated drain pads into real copper for current and heat spreading; do not neck them down right at the package.", "Keep each gate loop short, tight, and referenced to its own source return to reduce ringing and false turn-on.", "Assume the board copper sets the real current limit; check temperature rise on the actual ESC geometry, not only the datasheet headline current.", ], pins: { LOW_SIDE_SOURCE: "1", LOW_SIDE_GATE: "2", HIGH_SIDE_DRAIN: ["3", "8"], HIGH_SIDE_SOURCE: "4", HIGH_SIDE_GATE: "5", LOW_SIDE_DRAIN: ["6", "7"], }, }) const highSidePullDownBjt = definePart({ id: "BC847BLP_7", kicadSymbol: "Transistor_BJT:Q_NPN_BEC", footprint: "Package_TO_SOT_SMD:SOT-883", defaultValue: "BC847BLP-7", datasheet: " description: "45 V, 100 mA NPN small-signal transistor in SOT-883", designNotes: [ "Use this device only as the helper pull-down for the high-side P-gate nets; do not put motor or supply current through it as a power path element.", "In this topology, keep emitter at GND, collector on the P-gate net, and drive the base through a resistor from the MCU.", "Give the base a defined pulldown so the PMOS high side stays off while the MCU is in reset or high-impedance.", "With a 470R P-gate pull-up on a 1S rail, this transistor sinks about 9 mA at full turn-on, which is still well inside the BC847BLP-7's capability.", "Check the B-E-C pin order against the footprint before layout release; small BJTs are easy to rotate or mirror by accident.", ], pins: { BASE: "1", EMITTER: "2", COLLECTOR: "3", }, }) const yageoRc0201Datasheet = " const kemetMlccDatasheet = " const defineYageoRc0201 = (mpn: string, value: string) => definePart({ id: mpn, kicadSymbol: "Device:R", footprint: "Resistor_SMD:R_0201_0603Metric", defaultValue: value, datasheet: yageoRc0201Datasheet, description: `Yageo ${mpn} 0201 1% thick-film resistor`, fields: { Manufacturer: "Yageo", MPN: mpn, Tolerance: "1%", Power: "0.05W", }, designNotes: [ "This BOM is locked to an exact Yageo RC0201FR-07 1% resistor, not a generic 0201 placeholder.", "Keep the 1% series on the back-EMF divider path so thresholds stay predictable across temperature.", "Re-check pulse and dissipation stress if the battery domain or gate network changes.", ], pins: { A: "1", B: "2", }, }) const defineYageoRc0402 = (mpn: string, value: string) => definePart({ id: mpn, kicadSymbol: "Device:R", footprint: "Resistor_SMD:R_0402_1005Metric", defaultValue: value, datasheet: " description: `Yageo ${mpn} 0402 1% thick-film resistor`, fields: { Manufacturer: "Yageo", MPN: mpn, Tolerance: "1%", Power: "0.063W", }, designNotes: [ "Use 0402 here where the resistor sees non-trivial continuous dissipation on the switching rail.", "Do not silently shrink these positions back to 0201 without re-checking power and temperature margin.", ], pins: { A: "1", B: "2", }, }) const defineKemetMlcc = ({ mpn, value, footprint, description, voltage, dielectric, designNotes, }: { mpn: string value: string footprint: string description: string voltage: string dielectric: string designNotes: readonly string[] }) => definePart({ id: mpn, kicadSymbol: "Device:C", footprint, defaultValue: value, datasheet: kemetMlccDatasheet, description, fields: { Manufacturer: "KEMET", MPN: mpn, Dielectric: dielectric, "Rated Voltage": voltage, }, designNotes, pins: { POS: "1", NEG: "2", }, }) const resistor30r = defineYageoRc0201("RC0201FR-0730RL", "30R") const resistor470r0402 = defineYageoRc0402("RC0402FR-07470RL", "470R") const resistor1k = defineYageoRc0201("RC0201FR-071KL", "1k") const resistor10k = defineYageoRc0201("RC0201FR-0710KL", "10k") const resistor100k = defineYageoRc0201("RC0201FR-07100KL", "100k") const resistor12k = defineYageoRc0201("RC0201FR-0712KL", "12k") const resistor15k = defineYageoRc0201("RC0201FR-0715KL", "15k") const resistor33k = defineYageoRc0201("RC0201FR-0733KL", "33k") const swdVrefIsolationDiode = definePart({ id: "RB751CS40,315", kicadSymbol: "Device:D_Schottky", footprint: "tinybee:D_SOD-882", defaultValue: "RB751CS40,315", datasheet: " description: "40 V small-signal Schottky diode in SOD-882 for isolated SWD Vref sensing", designNotes: [ "This diode lets the programming header sense the target 3.3 V rail without back-powering the TLV755 when an external tool drives Vref.", "Place it close to the programming pads, and keep the +3V3 cathode side on the board rail with the isolated anode side only on the header Vref pin.", ], pins: { K: "1", A: "2", }, }) const capacitor100nF0402 = defineKemetMlcc({ mpn: "C0402C104K4RACTU", value: "100nF", footprint: "Capacitor_SMD:C_0402_1005Metric", description: "KEMET 100nF 16V X7R MLCC, 0402", voltage: "16V", dielectric: "X7R", designNotes: [ "This exact 16 V X7R part is locked for the 100 nF bypass positions in this 1S design.", "Use it for local high-frequency bypass on +BATT or V3.3; do not silently substitute a lower-voltage or poor-stability dielectric.", ], }) const capacitor100nF0201 = defineKemetMlcc({ mpn: "C0201C104K9PACTU", value: "100nF", footprint: "Capacitor_SMD:C_0201_0603Metric", description: "KEMET 100nF 6.3V X5R MLCC, 0201", voltage: "6.3V", dielectric: "X5R", designNotes: [ "Use this only for local low-voltage decoupling like the CH32 VDD bypass.", "Do not silently reuse this 0201 part on the battery rail; keep the battery-facing 100 nF positions on the 16 V 0402 part.", ], }) const capacitor1uF0402 = defineKemetMlcc({ mpn: "C0402C105K8PAC7411", value: "1uF", footprint: "Capacitor_SMD:C_0402_1005Metric", description: "KEMET 1uF 10V X5R MLCC, 0402", voltage: "10V", dielectric: "X5R", designNotes: [ "This exact 10 V X5R part is the TLV755 output capacitor and satisfies the regulator's 1 uF ceramic requirement.", "A 0402 body is acceptable here, but this is still a regulator output part, not a battery-domain bypass; do not shrink it further without checking bias derating and stability.", ], }) const capacitor22uF0805 = defineKemetMlcc({ mpn: "C0805C226M8PACTU", value: "22uF", footprint: "Capacitor_SMD:C_0805_2012Metric", description: "KEMET 22uF 10V X5R MLCC, 0805", voltage: "10V", dielectric: "X5R", designNotes: [ "This exact 10 V X5R 0805 part is the 1S battery-side bulk capacitor.", "Keep this input bulk capacitor in 0805 or larger; do not shrink it back to 0603 without re-checking effective capacitance at 1S bias.", ], }) const C1 = instantiate(capacitor1uF0402, "C1") const C2 = instantiate(capacitor100nF0201, "C2") const C3 = instantiate(capacitor100nF0402, "C3") const C4 = instantiate(capacitor22uF0805, "C4") const C5 = instantiate(capacitor100nF0201, "C5") const C6 = instantiate(capacitor22uF0805, "C6") const Q1 = instantiate(complementaryHalfBridge, "Q1") const Q2 = instantiate(complementaryHalfBridge, "Q2") const Q3 = instantiate(complementaryHalfBridge, "Q3") const Q4 = instantiate(highSidePullDownBjt, "Q4") const Q5 = instantiate(highSidePullDownBjt, "Q5") const Q6 = instantiate(highSidePullDownBjt, "Q6") const J_CTRL = instantiate(controlHeader, "J_CTRL") const J_BATT = instantiate(batteryHeader, "J_BATT") const J_MOTOR = instantiate(motorHeader, "J_MOTOR") const TP_PROG = instantiate(programmingHeader, "TP_PROG") const D1 = instantiate(swdVrefIsolationDiode, "D1") const R1 = instantiate(resistor30r, "R1") const R2 = instantiate(resistor1k, "R2") const R3 = instantiate(resistor10k, "R3") const R4 = instantiate(resistor470r0402, "R4") const R5 = instantiate(resistor30r, "R5") const R6 = instantiate(resistor1k, "R6") const R7 = instantiate(resistor10k, "R7") const R8 = instantiate(resistor470r0402, "R8") const R9 = instantiate(resistor30r, "R9") const R10 = instantiate(resistor1k, "R10") const R11 = instantiate(resistor10k, "R11") const R12 = instantiate(resistor470r0402, "R12") const R13 = instantiate(resistor15k, "R13") const R14 = instantiate(resistor33k, "R14") const R15 = instantiate(resistor100k, "R15") const R16 = instantiate(resistor15k, "R16") const R17 = instantiate(resistor33k, "R17") const R18 = instantiate(resistor100k, "R18") const R19 = instantiate(resistor15k, "R19") const R20 = instantiate(resistor33k, "R20") const R21 = instantiate(resistor100k, "R21") const R22 = instantiate(resistor100k, "R22") const R23 = instantiate(resistor100k, "R23") const R24 = instantiate(resistor100k, "R24") const R25 = instantiate(resistor12k, "R25") const R26 = instantiate(resistor33k, "R26") const R27 = instantiate(resistor1k, "R27") const U1 = instantiate(tlv75533pdqnt, "U1") const U2 = instantiate(ch32v203f8u6, "U2") const tinybeeEscChannel = defineCircuit({ name: "tinybee-esc-channel", source: "projects/tinybee/circuitd/tinybee-esc-channel.ts", description: "1S CH32-based tinybee ESC channel aligned to the openwch RISC-V_ESC V203 pinout", parts: [ C1, C2, C3, C4, C5, C6, D1, J_BATT, J_CTRL, J_MOTOR, Q1, Q2, Q3, Q4, Q5, Q6, R1, R2, R3, R4, R5, R6, R7, R8, R9, R10, R11, R12, R13, R14, R15, R16, R17, R18, R19, R20, R21, R22, R23, R24, R25, R26, R27, TP_PROG, U1, U2, ], nets: [ net( "+BATTERY", pin(J_BATT, "PIN1"), pin(C3, "POS"), pin(C4, "POS"), pin(C6, "POS"), pin(Q1, "HIGH_SIDE_SOURCE"), pin(Q2, "HIGH_SIDE_SOURCE"), pin(Q3, "HIGH_SIDE_SOURCE"), pin(R4, "B"), pin(R8, "B"), pin(R12, "B"), pin(R25, "B"), pin(U1, "IN"), pin(U1, "EN"), ), net("/ADC_VOLTAGE_SENSE", pin(C5, "POS"), pin(R25, "A"), pin(R26, "B"), pin(U2, "PA1/ADC1")), net("/A_HIGH_COMMAND", pin(R2, "B"), pin(U2, "PA10/CH3")), net("/A_LOW_COMMAND", pin(R1, "B"), pin(U2, "PB1/ADC9/OP1O1/CH3N")), net("/A_HIGH_CONTROL", pin(Q4, "BASE"), pin(R2, "A"), pin(R3, "B")), net("/A_LOW_GATE", pin(Q1, "LOW_SIDE_GATE"), pin(R1, "A"), pin(R22, "B")), net("/A_P_GATE", pin(Q1, "HIGH_SIDE_GATE"), pin(Q4, "COLLECTOR"), pin(R4, "A")), net("/A_BACK_EMF", pin(R13, "A"), pin(R14, "B"), pin(R15, "A"), pin(U2, "PA5/ADC5/OP2N1")), net("/B_BACK_EMF", pin(R16, "A"), pin(R17, "B"), pin(R18, "A"), pin(U2, "PB10/OP2N0")), net("/C_BACK_EMF", pin(R19, "A"), pin(R20, "B"), pin(R21, "A"), pin(U2, "PB11/OP1N0")), net("/BACK_EMF_COMMON", pin(R15, "B"), pin(R18, "B"), pin(R21, "B"), pin(U2, "PB14/OP2P0"), pin(U2, "PB15/OP1P0")), net("/B_HIGH_COMMAND", pin(R6, "B"), pin(U2, "PA9/CH2")), net("/B_LOW_COMMAND", pin(R5, "B"), pin(U2, "PB0/ADC8/OP1P1/CH2N")), net("/B_HIGH_CONTROL", pin(Q5, "BASE"), pin(R6, "A"), pin(R7, "B")), net("/B_LOW_GATE", pin(Q3, "LOW_SIDE_GATE"), pin(R5, "A"), pin(R23, "B")), net("/B_P_GATE", pin(Q3, "HIGH_SIDE_GATE"), pin(Q5, "COLLECTOR"), pin(R8, "A")), net("/C_HIGH_COMMAND", pin(R10, "B"), pin(U2, "PA8/CH1")), net("/C_LOW_COMMAND", pin(R9, "B"), pin(U2, "PA7/ADC7/OP2P1/CH1N")), net("/C_HIGH_CONTROL", pin(Q6, "BASE"), pin(R10, "A"), pin(R11, "B")), net("/C_LOW_GATE", pin(Q2, "LOW_SIDE_GATE"), pin(R9, "A"), pin(R24, "B")), net("/C_P_GATE", pin(Q2, "HIGH_SIDE_GATE"), pin(Q6, "COLLECTOR"), pin(R12, "A")), net("/PWM_INPUT", pin(J_CTRL, "PIN1"), pin(R27, "A")), net("/PWM_INPUT_MCU", pin(R27, "B"), pin(U2, "PA0/WKUP/ADC0")), net("/OPA_ZERO_CROSS_INTERRUPT", pin(U2, "PA2/ADC2/OP2O0"), pin(U2, "PA3/ADC3/OP1O0"), pin(U2, "PA4/ADC4/OP2O1")), net("/PHASE_A", pin(J_MOTOR, "PIN1"), pin(Q1, "HIGH_SIDE_DRAIN"), pin(Q1, "LOW_SIDE_DRAIN"), pin(R13, "B")), net("/PHASE_B", pin(J_MOTOR, "PIN2"), pin(Q3, "HIGH_SIDE_DRAIN"), pin(Q3, "LOW_SIDE_DRAIN"), pin(R16, "B")), net("/PHASE_C", pin(J_MOTOR, "PIN3"), pin(Q2, "HIGH_SIDE_DRAIN"), pin(Q2, "LOW_SIDE_DRAIN"), pin(R19, "B")), net("/SWD_CLOCK", pin(TP_PROG, "PIN1"), pin(U2, "PA14/SWC/PA11/UDM")), net("/SWD_DATA", pin(TP_PROG, "PIN2"), pin(U2, "PA13/SWD/PA12/UDP")), net("/SWD_VREF", pin(D1, "A"), pin(TP_PROG, "PIN4")), net( "GND", pin(J_BATT, "PIN2"), pin(C1, "NEG"), pin(C2, "NEG"), pin(C3, "NEG"), pin(C4, "NEG"), pin(C6, "NEG"), pin(Q1, "LOW_SIDE_SOURCE"), pin(Q2, "LOW_SIDE_SOURCE"), pin(Q3, "LOW_SIDE_SOURCE"), pin(Q4, "EMITTER"), pin(Q5, "EMITTER"), pin(Q6, "EMITTER"), pin(R3, "A"), pin(R7, "A"), pin(R11, "A"), pin(R14, "A"), pin(R17, "A"), pin(R20, "A"), pin(R22, "A"), pin(R23, "A"), pin(R24, "A"), pin(R26, "A"), pin(C5, "NEG"), pin(TP_PROG, "PIN3"), pin(U1, "GND"), pin(U1, "THERMAL_PAD"), pin(U2, "GND"), ), net("+3V3", pin(C1, "POS"), pin(C2, "POS"), pin(D1, "K"), pin(U1, "OUT"), pin(U2, "VDD")), net("unconnected-(U2-PA6{slash}ADC6{slash}OP1N1{slash}BKI-Pad20)", pin(U2, "PA6/ADC6/OP1N1/BKI")), ], }) const outputPath = path.resolve(__dirname, "generated", " const main = () => { mkdirSync(path.dirname(outputPath), { recursive: true }) writeFileSync(outputPath, exportKiCadNetlist(tinybeeEscChannel), "utf8") process.stdout.write(`${outputPath}\n`) } if (import.meta.main) { main() } export { tinybeeEscChannel } export default tinybeeEscChannel](https://image.24vids.com/tw-2033034293996359989/amplify_video_thumb/2033034066312695808/img/3d2AiTfCll6BSvuw.jpg)
![Just in $AMD Anush "Speed is the moat"|ROCm🎙️ In the race to define the future of AI, what's the one advantage that truly lasts? It's not proprietary tech, argues Anush Elangovan Elangovan, VP of AI Software at AMD , but the sustainable speed of innovation. He explains why AMD is rejecting the "walled garden" model for its open source ROCm stack, betting that an open community flywheel is the key to victory. Listen to understand how this open strategy is designed to out-innovate closed systems by empowering developers to solve everything from frontier-model challenges to the mundane, everyday problems that define the "last mile" of AI. AMD ROCm Software: Part 1 Transcript [00:00:00] Andrew Zigler: Joining me is Anush Elangovan, VP of AI software at AMD. And when people talk about AI compute, the conversation often stops at hardware specs, but it's more than just physical chips that win the game. It's also the software ecosystems supporting them. [00:00:18] Andrew Zigler: The prevailing strategy in the industry has been to build something like a walled garden. You know, something closed, proprietary locks, developers in. But AMD is betting on an entirely different play, open source acceleration, and with rock, their open source AI software stack. AMD is building not just hardware parity, but an innovation flywheel that's powered by the community with interoperability and the freedom to scale without all of that pesky lockin. [00:00:48] Andrew Zigler: And in this world, speed is your moat and how fast you can innovate while your platform remains open, flexible, and standardize across all of its applications. That's what we're gonna explore [00:01:00] today. So Anush, I'm really excited to have you here. Welcome to Dev Interrupted. [00:01:04] Anush Elangovan: Thanks for having me. Uh, super excited to chat about it. [00:01:07] Andrew Zigler: Amazing. Well, let's go ahead and dive right in with kind of what I laid it out with in the beginning, the idea of the moat and it being about speed. I wanna unpack that a bit because that came from you when you and I first spoke. And I, and I want to know, you know, how do you define speed inside of AMD beyond just things like hardware, benchmarks. [00:01:27] Anush Elangovan: Yeah, that's a very good question. So when we typically talk about speed, everyone's like, Hey, hardware benchmark specs, right? Like, uh, memory bandwidth or, or flops. And that is one important part of it, uh, AMD does very well. With that, we do have, a, a very good history of executing on that axis. [00:01:47] Anush Elangovan: But when I say speed is the moat, it is about, uh, how we prepare, how we build the muscle to run the race for a long time and run it fast. And it is [00:02:00] not about a single point in time that you've, you've beat some you know, benchmark and, and you declare victory. It's about building the ability to consistently develop and deliver. [00:02:13] Anush Elangovan: Both hardware and software innovation at scale and do it fast, right? Like, you know, we we're increasingly getting to a point where models come out and they're, uh, you know, a year or two ago it was like, Hey, they work on AMD on day zero, which is great, but now they are performing on AMD the day it releases, right? [00:02:32] Anush Elangovan: So, what does it take to Prefetch where the industry is going? Be prepared to intercept. At that point is what you know, I, I refer to as you know, the, the speed factor in, in creating this mode, right? And the mode is just shed all things that hold you back and run as fast as you can. [00:02:53] Anush Elangovan: Uh, because the pace of innovation that is, uh, being seen in, in AI [00:03:00] industries is just. Amazing. Right? And it's like, it's transformational at at how you generate electricity. It's transformational as at how you build data centers. It's transformational at how you deploy compute, networking. It's transformational at what kind of use cases you, you know, uh, use AI for. [00:03:17] Anush Elangovan: Uh, and for that, you need to be prepared to, see what comes tomorrow and be prepared to run the race tomorrow. [00:03:23] Andrew Zigler: Yeah, it's a really great perspective because it highlights that it's not just like a checkpoint that you run through. I like how you called out, like it's not just hitting that benchmark or being the best in class at that moment, in that snapshot, it's about having a. The throughput and about having that dedication to the idea and continuing to deliver on it. [00:03:43] Andrew Zigler: It's not just crossing the threshold, but it's also being the engine. And that's what, that's what protects a business. That is the moat, because the moat is that innovation layer, the faster and more, uh, future forward. That you can work and think, [00:04:00] you know, the better. Uh, we, we talk a lot about like future forward work styles. [00:04:04] Andrew Zigler: Like what are the things I could be doing right now today that are gonna be like, way more useful tomorrow? Let, let's abandon those, workflows that are older and that kind of like, that translates into. An advantage when you work that way. You know, what kind of things have you learned working with, uh, like across all spectrums of people who would use ROCm, right? [00:04:23] Andrew Zigler: You have like the developers, but then you also have the enterprises and you have this large span of adoptees, right? So what is the, what does that look like that you learn? [00:04:32] Anush Elangovan: Yeah, so, so the way I look at it is there are gonna be pockets of different, uh, you know, cadences, right? Like, so people who are deploying in enterprises, for example, right? The validation and how long it takes for them to deploy an LLM that's secure. It's, with guardrails, et cetera, maybe longer. [00:04:52] Anush Elangovan: but you still have to go through the process and you have to be prepared to like, walk that walk to deploy an enterprises. That doesn't mean it's [00:05:00] not fast, that's as fast as you can do for that industry, right? And if you are deploying AI in healthcare, right, it's, it's got its own, uh, cycle. [00:05:07] Anush Elangovan: but in each one of these, you want to see how, like, go down to the essence of what is it that you actually have to do. And, you know, I, I, I like how you framed it. It's like it's, you shed your prior assumptions of how things are done, right. And, and you kind of build up from a, uh, first principles, uh, approach to say, this is how I could use AI to unlock, whatever I'm doing. [00:05:33] Anush Elangovan: And, and, some of it, you know, it's good to really step back and look at. Just question every part of it, right? Like right now you're getting chat GPT and, Gemini competing for like, math, olympiads and, and, uh, college, uh, reasoning, uh, tests. Right? And, and those are like that, that is amazing and increasingly like complex tasks that they're trying to do. [00:05:58] Anush Elangovan: But there may also be like. [00:06:00] More mundane things that AI could, could get applied to. Right? And, and so when we think about shedding old ways, you wanna shed it not just in like the tip of the spear. It's like, you know, I'm gonna see what's the frontier model. It's also, it could be something as simple as. [00:06:18] Anush Elangovan: How do you choose a, a movie, uh, you know, like a recommendation system, right? Or, or, uh, an automated, uh, flight, uh, rebooking system. So the moment, you know, your flight is late, uh, right now it's a notification, right? It's like, oh, you got a text message saying your flight's late. And I got that like three times this week. [00:06:38] Anush Elangovan: But anyway, uh, and, and, and, and, I was just like, okay, so if I were to rethink this. All this MCPs that we have that should be hooked up into an MCP that says, your flight's delayed. Here are your options. If you want, you know, these are the paid options. Yeah. Here are the free options. This will get you back into your you know, Toronto airport [00:07:00] tonight. [00:07:00] Anush Elangovan: Or if you stay, here's a hotel plus this, plus this, plus. It's just like, go ahead is all I should say. Versus now I'm like, okay, can someone, you know, can I call a travel agent? Can I do this? Can I go online and log into And you know, so we gotta fundamentally rethink even those like small, nuances of, things that we do that can be automated out and AI is really, really good at doing something like this, right? Maybe I just explained an AI startup idea right now. Somebody should just start that. [00:07:29] Andrew Zigler: I think you did. Yeah, you definitely did. Someone, one of our listeners is definitely going to lift that off of you. I, I, I, you know, I hate being on the receiving end of those. You feel a little helpless and then you have to like, follow the whole flow. So I know what you mean. Like I, I like how you called out that the build and this like. [00:07:45] Andrew Zigler: Where speed is your moat and the innovation layer is protecting you, is what makes you better than your competitors. How you scale that and you bring that to market. So by understanding the problems that you're solving, uh, throwing away those older assumptions, but also [00:08:00] recognizing that like. We're building every single day, new things and new ways of using stuff that we're still figuring out the implications of. [00:08:08] Andrew Zigler: And so when you have a lot of velocity and you're introducing a lot of new ideas, and maybe you have that workflow now that automatically rebook your flight off of your late flight text message, and uh, I know I would certainly use it, but you know, what kind of philosophies guide the way that y'all think about building this ecosystem to manage that stability while letting folks. [00:08:29] Andrew Zigler: Play with the speed and the assumptions and the airplane re bookings. [00:08:34] Anush Elangovan: so, so I think, you know, we need to peel one layer down, right? and the philosophy is, Hey, we, we just discovered electricity, right? And you know what we're gonna do? We are gonna make motors, uh, or dynamos, right? Like engines. Uh, sure. We don't know if it's gonna be a Ferrari that you're gonna make, or it's a a a a dump truck. [00:08:57] Anush Elangovan: That's good for doing this. But let's [00:09:00] let, which is also required, right? You need a dump truck. You need a garbage truck. And, [00:09:04] Andrew Zigler: Yeah. You need the [00:09:04] Anush Elangovan: course you need, uh, a Ferrari for a midlife crisis, right? So, [00:09:09] Andrew Zigler: precisely. [00:09:10] Anush Elangovan: But, but my, uh, point is what do we build next? And, uh, and this is what I meant by like, okay, let's, let's take those baby steps to build the. [00:09:20] Anush Elangovan: Infrastructure that's required that we know we'll have to use, right? So, so if I just discovered electricity, okay, great. Now one, how do I save this electricity and how do I use it? So there's battery technology, so you need to do something like that, right? Like so. But then you also want to make it into an actionable thing. [00:09:37] Anush Elangovan: You want to make it for like automobiles, or you wanna use it for, you know, powering, uh, entire cities. So it is that transformational. So, uh, AI is that transformational. So, if you distill down, it'll, it'll come down to how do we think about, what we can do with this this fundamental technology that, We may not be aware of what it [00:10:00] is gonna unlock next, but at least you know the next step is clear, right? It's like a dense fog, you know, it's gonna be like, it, it's the right path. You see the light, but it's kind of like out there and, and the steps you're taking are concrete and you're like, okay, this is good. [00:10:16] Anush Elangovan: I, this is better than where I was or where we were. So we are moving forward. So you can build with the. Intuition from what you see in the short term and a tactical view, but towards what you think the future is gonna be. [00:10:28] Andrew Zigler: Right. You almost like we're all in this like fog of war, right? And like you said, you're reaching out and you're trying to step through it. You could think of it too, as like you're in the dark and your hands are up in front of you and you know that. You're, you're not gonna run your face into a wall because your hands are out in front of you, but you're not gonna maybe do much better than that. [00:10:45] Andrew Zigler: So that's kind of like, I think the eco, the, the industry, the world that we find ourselves in, uh, and we all have to, then this becomes the power of an ecosystem, of a group of people working together to create that layer of, [00:11:00] uh, of establishing the [00:11:01] Anush Elangovan: exactly. And I, I, I just, instead of, you know, saying fog of war I describe it as like, you're in this. Beautiful valley with like a morning, uh, fog that's in. You can smell the flowers. You, you hear the birds. You are like, okay, it's, we are in like, uh, utopian paradise and yes, I just need to like, continue the walk, right? [00:11:24] Anush Elangovan: and then move forward with that, conviction that you're in the right spot. [00:11:27] Andrew Zigler: Yeah. So let's talk about that ecosystem world. This nice, I love how you describe it, this grassy side of a hill in the morning that's covered in some mist and maybe we can't see 30 feet in one direction, but it sure is a beautiful hill and it smells nice. And so we're all here. And why is, in that world, why is. [00:11:44] Andrew Zigler: You know, open source, their strategic advantage that y'all are going for in the AI hardware market. And, and then how does like ROCm turn that into wins for people within that ecosystem? [00:11:56] Anush Elangovan: you know, the, the way we look at it is this, is kind of like how I view [00:12:00] AI and the ecosystem, right? But, but it is for everyone to enjoy. Uh, and so we do want to make sure that. You know, it is, uh, beneficial for everyone. [00:12:09] Anush Elangovan: The ecosystem can come in and, and innovate. It's an open innovation engine. and uh, it is very different from, you know, having a walled garden with, Hey, only I know how to do this and I'm gonna do it and throw it over the fence and you can use it or keep walking, right? So we'd like to be good citizens that way, but also. [00:12:30] Anush Elangovan: Uh, it is self-fulfilling in a way, right? Like it, the, the pace at which we innovate with open source is unmatched. Like, you know, our serving engines are like VLLM and, and sg l. Those things, uh, those frameworks are like super, super aggressive in terms of how fast they come out with features and how fast they can you know, get performant models out. [00:12:52] Anush Elangovan: And that compared with what, uh, you'd get from, you know, the likes of like T-R-T-L-L-M or something is always lagging, right? Because you [00:13:00] just can't keep up with you know, 200 commits a week just on one particular model to get that model really performant [00:13:06] Andrew Zigler: And, and, and in that world where, you know, everyone can enjoy the winds of this, what kind of customer stories or innovation stories have really stood out to you and excite you about building and creating this place for developers? [00:13:19] Anush Elangovan: Yeah. So I think the parts that are super exciting for me are when when we get to see a customer that is first skeptical. Then they start a little like, okay, fine, we'll give you a chance. Uh, we do a simple, uh, POC and then they're like, huh, this seems to work. Yeah, we told you it works. [00:13:42] Anush Elangovan: You don't have to change one line of code. Really? Yes, no need to change one line of code. Okay, let's try a production workload. So then they try it. Oh, you're more performant than the competition. Yes. We're more performant than, than the competition. So how much does it cost? And we're like, oh, it's your TCO is better with, uh, [00:14:00] AMD. [00:14:00] Anush Elangovan: So again, they're like, wow, okay, good. So now how do we deploy at scale? And then we go deploy it at scale. And when they give a thumbs up on that and they say, this is good, right? That's when you know, you, you see it go full circle from like, oh, we, we've never heard about AMD to like actually deploy to tens of thousands of GPUs In the order of a few months, right? It, it, it really is fascinating to see and very exciting and invigorating to [00:14:28] Andrew Zigler: Yeah. At like a great exposure to a lot of interesting problems. And, and then people using the infrastructure, the, the technology available to solve those problems. Really specific problems by the way, that's often why they're bringing their data and AI to it, uh, is because it is really specific and important for them. [00:14:45] Andrew Zigler: And there's a, a lot I think that other engineering orgs can learn and even emulate from AMD's success and, and having this open source ecosystem and it causing this acceleration within. You [00:15:00] know, uh, customers and enterprises that use and adopt the tools and, and, and that creates an advantage. And that goes back to why we're talking and like the real thesis of our conversation today. [00:15:10] Andrew Zigler: So how do you think engineering leaders that are listening to this and obviously tapping into this great success AMD has from an open source flywheel, how do you think other, other folks building in the same space can foster that open, first, that open source oriented culture in order to, you know, accelerate their innovation goals? [00:15:29] Anush Elangovan: Yeah, that's a very good question. So the startup that um, was acquired by AMD we, we built, I mean, we started off doing iot stuff and you know, smart ring and all that, right? But in the, the end of like, uh, and not the end, the last six years of the company was building ML compilers. [00:15:47] Anush Elangovan: And ml, ML compilers are like super, uh, complicated, sophisticated, advanced algorithms, dah, dah, dah. but it was all open source, right? So our VCs were like, wait, what do you mean your core [00:16:00] IP is open source? And um, the speed is the moat applied even then, right? It was just like, yes, if you have an idea that. [00:16:08] Anush Elangovan: Because someone saw this idea that you are, they're gonna be able to catch up, then you probably have the wrong idea anyway. But if they are, you know, you execute and they're gonna catch up, that you should assume they're gonna catch up. Right? So you gotta move forward. So keeping it open source is super important. [00:16:25] Anush Elangovan: But also to your question on like, you know, the learnings from an AMD standpoint, right? If there are, hard problems, I'd say dig in and work through it, right? Like there's no way but through it, right? That should be the simple mentality. And more, uh, frequently than not. you'll see that you'll just make it through in a, in, in good form. [00:16:52] Anush Elangovan: But if you doubt it and you're like, oh, I don't know if I should commit, if I'm, I, you know, what should just commit to do the right thing [00:17:00] every step, right? Every step, and just keep taking one step in front of the other. And in no time you'll see that you'll be running. Right. And, and yes, the first few steps will be like, yeah, everyone's complaining about your software quality. [00:17:15] Anush Elangovan: Everyone's complaining about this and that, and it doesn't work. And, and a few steps in, you know, you get, you get the hang of all the complaints that are coming in. You get the feedback loop. You're like, okay, what, what are you prioritizing again? One step in front of the other, right? You just keep knocking that out and then you get to a point where you're, it just becomes second nature, right? To do the, to do the right thing. And, and then yes, if someone gives you two options, you'll be like, fine. This is, uh, you know, there's always the resource trade off. There's always a human capital trade off, but what's the right thing to do? of course, I, I'm pragmatic about what we choose, but, but if the right thing for your long-term success is dig in, go first, principles, make it [00:18:00] happen. [00:18:00] Anush Elangovan: Well. Then just go for that. There's, there is no shortcut to [00:18:04] Andrew Zigler: acknowledging, you know, how it aligns with your mission, your core company goals, and what you're looking to achieve. And, and I, I love how you rightfully called out that in the open source world and you know, you have your technology that you've built, what you think is your moat upon, right? [00:18:22] Andrew Zigler: It's your code and, and to open source that, or to just make it where anyone could peer in is, you know. Scary in one regard, but two, it just kind of feels like you're handing away your throne room in some kind of sense, a very direct feeling sense. But the ultimately, you were really right to call out, and this is something I think about all the time, that the real power there is still the speed This the speed. [00:18:42] Andrew Zigler: That was the moat at the beginning of our conversation. It's the speed in combination with your. Very specific domain understanding of what you're building and what you're creating, and your new role as the steward of that world and how people plug into it, which [00:19:00] has frankly, a lot more influence and power than lording over a closed. [00:19:04] Andrew Zigler: You know, repository or an ecosystem, and like you said, like throwing things over the wall. Sure. There, there might be people always on the other side of that wall, but you're not gonna have a great connection with them. You're not gonna be able to really clearly understand them. I, I like your metaphor of the side of the field of the mountain a lot more. [00:19:23] Andrew Zigler: But, but in the, in this world, you know, where. That speed is, is the power and, and open source is just one way that you can harness that speed to get really far ahead and to innovate. , There's other parts of this equation that you can be experimenting with too, and I'd love to pick your brain about them as a software leader and, and, and one of them is about looking forward and kind of understanding that future that we're all building towards and beyond today's models and hardware. [00:19:48] Andrew Zigler: You know, what do you see as the next major bottleneck or opportunity in the AI compute space? As, as you know, enterprises and folks start to get a little more mature about what's available to [00:20:00] them. [00:20:00] Anush Elangovan: Yeah, I think, the bottleneck and opportunity is, uh, what I'd call, call walking the last mile of ai. Right. Uh, and like I I, I gave you an example, uh, previously, but, but it's similar to that. It's like there are cases where Humans have so many, uh, things to do in your day. You know, like the, if we sit down and actually had a customer focus like, okay, these customers lives, I'm gonna save four hours of this customer's life. And if you actually sit down and look at all of that, it'll be. Easily automatable, easily you know, uh, applicable, uh, for ai, right? [00:20:39] Anush Elangovan: Like, but then making it happen is gonna take a little bit, right? It's like maybe it's, uh, paying your utility bill, right? Or something like that, right? Or, or, your healthcare explanation of benefits. Uh, like, I'm sure you get an explanation of benefits, and I'm like, I, I don't even know what that thing is. [00:20:55] Anush Elangovan: It's just like EOB and like. [00:20:57] Andrew Zigler: it's a big, a big old PDF. Yeah, [00:21:00] exactly. [00:21:01] Anush Elangovan: Like, like, I'm like great straight to the, uh, shredder, right? And but that could be, you know, automated with the ai, right? It, it, it'd be like, Hey, the summary of this thing is you went and visited this day. Everything is okay. Everything is paid for, so don't worry, it's not a bill. [00:21:17] Anush Elangovan: That again, the same, uh, thing, but the sense of what that information overload is could be. Digested by ai, uh, accumulated over time and retrieved when you need it. Like, I don't, I actually don't even need to know this EOB right now, unless of course, whenever I need to know it, that maybe, you know, like for some benefits I need to figure out what do, what did I do over the past year and how do I apply it? Source:](https://image.24vids.com/tw-1994972575768727743/amplify_video_thumb/1994972220930621440/img/_5CCZJyz89oDfqQJ.jpg)
