5 ms·

Intel doesn't seem to be able to execute. It's not just pumping out GPUs - for AI you need drivers, and the equivilent of CUDA and all the various libraries bui

Intel doesn't seem to be able to execute. It's not just pumping out GPUs - for AI you need drivers, and the equivilent of CUDA and all the various libraries built on CUDA like cuDNN. They do have OneAPI but it hasn't caught on like CUDA in that space. It's kind of too bad since OneAPI is open and CUDA is not.

HDThoreaun 3y ago
Right but the market is saying that a dominant GPU business is worth more than a trillion dollars. Just hard for me to believe that they can't get the business off the ground with that kind money on the table. Can't they just hire all of nvidia's developers and pay them 5x as much?
- tgma 3y ago
  - Can't they just hire all of nvidia's developers and pay them 5x as much? No.
- UncleOxidant 3y ago
  > Can't they just hire all of nvidia's developers and pay them 5x as much? Lol... Intel is famously stingy when it comes to salaries.
  - HDThoreaun 3y ago
    for a trillion dollars though... eventually you have to believe Pat gets fired and replaced by someone who is 100% all in on GPUs if he can't figure this out
    - UncleOxidant 3y ago
      I'd argue that Intel being stingy with salaries is a big part of why they're so behind here. They just don't seem to be very serious about this. Intel has made several runs at the GPU market over the years and they just keep ending up where they are. And now NVDA has such a huge advantage (software and hardware) that it just gets harder and harder (and more expensive) to overcome. Probably Intel's best bet now would be to try to be the fab for NVDA.
    - orzig 3y ago
      Maybe everything changes at $1 trillion, but I definitely see smaller (but public) companies leaving money on the table because it would require cultural change.
  - i_have_an_idea 3y ago
    You have no idea. There are a lot of senior engineers at NVDA making 7 figures total comp annually. How many are there at Intel?
    - paulmd 3y ago
      Intel has senior engineers who received two and three digit retention bonuses this year, on top of the 20% pay cut. Principals usually broke into the three digit range from what I hear.
- andrepd 3y ago
  The market is also saying that Tesla is worth more than BMW, VW, Audi, Mercedes, Toyota, Hyundai, Fiat, Ford, and dozens of others combined. Mehh, I don't know.
  - UncleOxidant 3y ago
    Exactly, the market isn't always rational. There's a lot of work on quantization in neural nets, for example, that can allow them to work sufficiently well on less capable hardware. It could be that some breakthrough there would obviate the need for NVDA hardware (or at least reduce how many are needed).
    - lotsofpulp 3y ago
      It is rational if you interpret market capitalization and share price movements as “the market is saying Tesla WILL be worth more than x,y z combined between time now and time whenever you want to sell it.” For different people, the timespan between now and when they may want or need to sell it is different, and thus different people will arrive at different conclusions. And note that “worth more” above simply means growth in market capitalization, so as long as someone is willing to buy the shares at a price supporting that increased market cap, then it does not matter if Tesla is still selling fewer cars than the others combined.
  - scrlk 3y ago
    "In the short run, the market is a voting machine but in the long run, it is a weighing machine."
    - TheAlchemist 3y ago
      Exactly this ! And as somebody with a significant short position, I would add - "The market can stay irrational longer than you can stay solvent" ! Their numbers for this and next Q are absolutely amazing. It's also quite "refreshing" - a company with great product, almost without competition (so far - it will come real quick). And fun part being their main advantage is probably CUDA and not even the chips itself (which by the way they don't manufacture - they "only" do the design). But still - even with those numbers, and even with this pace of growth (both being absolutely not sustainable, and will probably reverse hard next year) - the valuation doesn't make any sense, especially given the current interest rates.
- nemothekid 3y ago
  >Can't they just hire all of nvidia's developers and pay them 5x as much? As time goes on I don’t see how you break the CUDA moat even if you had all of nvidia Al’s engineers. CUDA means you need everyone in AI to target your new (hopefully open) platform and that platform is faster than CUDA is. Given how most frameworks of the last 10 years have been optimized for CUDA you would need to turn around a global sized cruise ship. If Intel’s GPUs are only 3% faster, will that be enough to rewrite my entire software stack for something not CUDA? If intel opts for a translation layer, could they ever match nvidia’s performance?
  - HDThoreaun 3y ago
    Well I'm not super experienced with GPU development but aren't most people using packages built on top of CUDA like pytroch etc? Would it be impossible to throw tons of resources at those packages so they handle whatever intel comes up with as well as they handle CUDA? If Intel is 10% slower but 50% cheaper and the open source stack you use has been heavily updated to work well with Intel drivers would that not be an enticing product?
    - UncleOxidant 3y ago
      Intel's been trying this for several years now (OneAPI and OpenVINO), but so far they haven't gotten the traction. CUDA is just really entrenched at this point.
    - michaelt 3y ago
      > aren't most people using packages built on top of CUDA like pytorch etc? Yes, and in fact both AMD and Intel have libraries. You can run Stable Diffusion and suchlike on AMD GPUs today, apparently. And you can export models from most ML frameworks to run in the browser, on phones and suchlike. > If Intel is 10% slower but 50% cheaper [...] would that not be an enticing product? Sometimes, yes. Some of the largest models apparently cost $600,000 in compute time to train [1], so halving that would be pretty appealing. However, part of the reason for nvidia's dominance is that if you're hiring an ML engineer for $160,000/year spending $1,600 to give them an RTX 4090 is chump change. [1] https://twitter.com/emostaque/status/1563870674111832066 https://twitter.com/emostaque/status/1563870674111832066
Mountain_Skies 3y ago
Over the past decade Intel seems to have become more interested in social causes than in technology, maybe with a side of government backrubbing to keep some income flowing.
- UncleOxidant 3y ago
  Nah, the biggest problem is that Intel became very risk averse. Yeah, they'll talk a good game on taking risks, but when it comes down to it people who took risks that failed tend to not be at Intel and other employees see that and think that maybe they need to play it safe.
  - johnvanommen 3y ago
    > Yeah, they'll talk a good game on taking risks, but when it comes down to it people who took risks that failed tend to not be at Intel and other employees see that and think that maybe they need to play it safe. I worked at Sears corporate when Amazon was getting big, about 25 years ago. Always made me chuckle when armchair quarterbacks on TV would wonder why Sears couldn't do what Amazon did. Bezos took tremendous risks in the late 90s and early 00s, while Sears was trying to figure out how to wring a few more pennies out of their stores. Sears Corporate was 110% focused on taking the existing business and maximizing profits, not on innovation of any kind whatsoever.
highwaylights 3y ago
I can really see Intel figuring this out. A lot of people on HN talking about Intel as an also-ran just like they spoke about AMD before Zen. Raptor Lake is at 7nm and incredibly competitive there (~2700 single core on geekbench, taken with a pinch of salt). They’re still planning on being on 1.8nm/18A within 2 years, while at the same time ramping up their GPU efforts (albeit using TSMC for 4nm). Nvidia is very much in the lead, but this is just the beginning. tldr; I ain’t hear no bell.
- andromeduck 3y ago
  The problem with Intel is: 1. They don't pay - Nvidia/Google/Apple easily pays 1.5-2x Intel before appreciattion. 2. They're cheap/beaurcratic. The office sucks, your laptop sucks. 3. They suck at software. https://pharr.org/matt/blog/2018/04/18/ispc-origins https://pharr.org/matt/blog/2018/04/18/ispc-origins 4. They can't develop/retain talent. Half the ML-HW/FW teams at AMD/Google/Nvidia/Apple are ex-Intel.
  - Panzer04 3y ago
    How is that indicative of not developing talent? Sounds like they have a fairly robust pipeline if so many end up working at other companies... (Retention notwithstanding :P)
    - andromeduck 3y ago
      They lost most of their talent in the early/mid 2010s but you're right to an extent for non-software. For software they're not a terrible place to accumulate domain specific knowledge but don't really retain people long enough or promote quick enough to make a difference.
- jjpprrrr 3y ago
  Raptor Lake uses the "Intel 7" node which is actually 10nm, not 7 nm. It does have roughly the same density as TSMC's 7nm node.
  - 3y ago

HDThoreaun 3y ago

Right but the market is saying that a dominant GPU business is worth more than a trillion dollars. Just hard for me to believe that they can't get the business off the ground with that kind money on the table. Can't they just hire all of nvidia's developers and pay them 5x as much?

tgma 3y ago
- Can't they just hire all of nvidia's developers and pay them 5x as much? No.
UncleOxidant 3y ago
> Can't they just hire all of nvidia's developers and pay them 5x as much? Lol... Intel is famously stingy when it comes to salaries.
- HDThoreaun 3y ago
  for a trillion dollars though... eventually you have to believe Pat gets fired and replaced by someone who is 100% all in on GPUs if he can't figure this out
  - UncleOxidant 3y ago
    I'd argue that Intel being stingy with salaries is a big part of why they're so behind here. They just don't seem to be very serious about this. Intel has made several runs at the GPU market over the years and they just keep ending up where they are. And now NVDA has such a huge advantage (software and hardware) that it just gets harder and harder (and more expensive) to overcome. Probably Intel's best bet now would be to try to be the fab for NVDA.
  - orzig 3y ago
    Maybe everything changes at $1 trillion, but I definitely see smaller (but public) companies leaving money on the table because it would require cultural change.
- i_have_an_idea 3y ago
  You have no idea. There are a lot of senior engineers at NVDA making 7 figures total comp annually. How many are there at Intel?
  - paulmd 3y ago
    Intel has senior engineers who received two and three digit retention bonuses this year, on top of the 20% pay cut. Principals usually broke into the three digit range from what I hear.
andrepd 3y ago
The market is also saying that Tesla is worth more than BMW, VW, Audi, Mercedes, Toyota, Hyundai, Fiat, Ford, and dozens of others combined. Mehh, I don't know.
- UncleOxidant 3y ago
  Exactly, the market isn't always rational. There's a lot of work on quantization in neural nets, for example, that can allow them to work sufficiently well on less capable hardware. It could be that some breakthrough there would obviate the need for NVDA hardware (or at least reduce how many are needed).
  - lotsofpulp 3y ago
    It is rational if you interpret market capitalization and share price movements as “the market is saying Tesla WILL be worth more than x,y z combined between time now and time whenever you want to sell it.” For different people, the timespan between now and when they may want or need to sell it is different, and thus different people will arrive at different conclusions. And note that “worth more” above simply means growth in market capitalization, so as long as someone is willing to buy the shares at a price supporting that increased market cap, then it does not matter if Tesla is still selling fewer cars than the others combined.
- scrlk 3y ago
  "In the short run, the market is a voting machine but in the long run, it is a weighing machine."
  - TheAlchemist 3y ago
    Exactly this ! And as somebody with a significant short position, I would add - "The market can stay irrational longer than you can stay solvent" ! Their numbers for this and next Q are absolutely amazing. It's also quite "refreshing" - a company with great product, almost without competition (so far - it will come real quick). And fun part being their main advantage is probably CUDA and not even the chips itself (which by the way they don't manufacture - they "only" do the design). But still - even with those numbers, and even with this pace of growth (both being absolutely not sustainable, and will probably reverse hard next year) - the valuation doesn't make any sense, especially given the current interest rates.
nemothekid 3y ago
>Can't they just hire all of nvidia's developers and pay them 5x as much? As time goes on I don’t see how you break the CUDA moat even if you had all of nvidia Al’s engineers. CUDA means you need everyone in AI to target your new (hopefully open) platform and that platform is faster than CUDA is. Given how most frameworks of the last 10 years have been optimized for CUDA you would need to turn around a global sized cruise ship. If Intel’s GPUs are only 3% faster, will that be enough to rewrite my entire software stack for something not CUDA? If intel opts for a translation layer, could they ever match nvidia’s performance?
- HDThoreaun 3y ago
  Well I'm not super experienced with GPU development but aren't most people using packages built on top of CUDA like pytroch etc? Would it be impossible to throw tons of resources at those packages so they handle whatever intel comes up with as well as they handle CUDA? If Intel is 10% slower but 50% cheaper and the open source stack you use has been heavily updated to work well with Intel drivers would that not be an enticing product?
  - UncleOxidant 3y ago
    Intel's been trying this for several years now (OneAPI and OpenVINO), but so far they haven't gotten the traction. CUDA is just really entrenched at this point.
  - michaelt 3y ago
    > aren't most people using packages built on top of CUDA like pytorch etc? Yes, and in fact both AMD and Intel have libraries. You can run Stable Diffusion and suchlike on AMD GPUs today, apparently. And you can export models from most ML frameworks to run in the browser, on phones and suchlike. > If Intel is 10% slower but 50% cheaper [...] would that not be an enticing product? Sometimes, yes. Some of the largest models apparently cost $600,000 in compute time to train [1], so halving that would be pretty appealing. However, part of the reason for nvidia's dominance is that if you're hiring an ML engineer for $160,000/year spending $1,600 to give them an RTX 4090 is chump change. [1] https://twitter.com/emostaque/status/1563870674111832066 https://twitter.com/emostaque/status/1563870674111832066

Mountain_Skies 3y ago

Over the past decade Intel seems to have become more interested in social causes than in technology, maybe with a side of government backrubbing to keep some income flowing.

UncleOxidant 3y ago
Nah, the biggest problem is that Intel became very risk averse. Yeah, they'll talk a good game on taking risks, but when it comes down to it people who took risks that failed tend to not be at Intel and other employees see that and think that maybe they need to play it safe.
- johnvanommen 3y ago
  > Yeah, they'll talk a good game on taking risks, but when it comes down to it people who took risks that failed tend to not be at Intel and other employees see that and think that maybe they need to play it safe. I worked at Sears corporate when Amazon was getting big, about 25 years ago. Always made me chuckle when armchair quarterbacks on TV would wonder why Sears couldn't do what Amazon did. Bezos took tremendous risks in the late 90s and early 00s, while Sears was trying to figure out how to wring a few more pennies out of their stores. Sears Corporate was 110% focused on taking the existing business and maximizing profits, not on innovation of any kind whatsoever.

highwaylights 3y ago

I can really see Intel figuring this out. A lot of people on HN talking about Intel as an also-ran just like they spoke about AMD before Zen. Raptor Lake is at 7nm and incredibly competitive there (~2700 single core on geekbench, taken with a pinch of salt). They’re still planning on being on 1.8nm/18A within 2 years, while at the same time ramping up their GPU efforts (albeit using TSMC for 4nm). Nvidia is very much in the lead, but this is just the beginning. tldr; I ain’t hear no bell.

andromeduck 3y ago
The problem with Intel is: 1. They don't pay - Nvidia/Google/Apple easily pays 1.5-2x Intel before appreciattion. 2. They're cheap/beaurcratic. The office sucks, your laptop sucks. 3. They suck at software. https://pharr.org/matt/blog/2018/04/18/ispc-origins https://pharr.org/matt/blog/2018/04/18/ispc-origins 4. They can't develop/retain talent. Half the ML-HW/FW teams at AMD/Google/Nvidia/Apple are ex-Intel.
- Panzer04 3y ago
  How is that indicative of not developing talent? Sounds like they have a fairly robust pipeline if so many end up working at other companies... (Retention notwithstanding :P)
  - andromeduck 3y ago
    They lost most of their talent in the early/mid 2010s but you're right to an extent for non-software. For software they're not a terrible place to accumulate domain specific knowledge but don't really retain people long enough or promote quick enough to make a difference.
jjpprrrr 3y ago
Raptor Lake uses the "Intel 7" node which is actually 10nm, not 7 nm. It does have roughly the same density as TSMC's 7nm node.
- 3y ago