Comment by singhrac

Comment by singhrac 3 days ago

I want to argue that graphics cards are really 3 markets: integrated, gaming (dedicated), and compute. Not only do these have different hardware (fixed function, ray tracing cores, etc.) but also different programming and (importantly) distribution models. NVIDIA went from 2 to 3. Intel went from 1 to 2, and bought 3 (trying to merge). AMD started with 2 and went to 1 (around Llano) and attempted the same thing as NVIDIA via GCN (please correct me if I'm wrong).

My understanding is that the reason is that the real market for 3 (GPUs for compute) didn't show up until very late, so AMD's GCN bet didn't pay off. Even in 2021, NVIDIA's revenue from gaming was above data center revenue (a segment they basically had no competition in, and 100% of their revenue was from CUDA). AMD meanwhile won the battle for Playstation and Xbox consoles, and was executing a turnaround in data centers with EPYC and CPUs (with Zen). So my guess as to why they might have underinvested is basically: for much of the 2010s they were just trying to survive, so they focused on battles they could win that would bring them revenue.

This high level prioritization would explain a lot of "misexecution", e.g. if they underhired for ROCm, or prioritized APU SDK experience over data center, their testing philosophy ("does this game work ok? great").

brudgers 3 days ago

The market segmentation you describe makes a lot of sense to me. But I don't think the situation is a matter of under-investment and is instead just fundamental market economics.

Nvidia can afford to develop a comprehensive software platform for the compute market segment because it has a comprehensive share of that segment. AMD cannot afford it because it does not have the market share.

Or to put it another way, I assume that AMD's efforts are motivated rational economic behavior and it has not been economically rational to compete heavily with Nvidia in the compute segment.

AMD was able to buy ATI because ATI could not compete with Nvidia. So AMD's graphics business started out trailing Nvidia. AMD has had a viable graphics strategy without trying to beat Nvidia...which makes sense since the traditional opponent is Intel and the ATI purchase has allowed AMD to compete with them pretty well.

Finally, most of the call for AMD to develop a CUDA alternative is based on a desire for cheaper compute. That's not a good business venture to invest in against a dominate player because price sensitive customers are poor customers.

Reply View 5 replies

quickthrowman 3 days ago

> Finally, most of the call for AMD to develop a CUDA alternative is based on a desire for cheaper compute. That's not a good business venture to invest in against a dominate player because price sensitive customers are poor customers.
Nvidia’s gross margins are 80% on compute GPUs, that is excessive and likely higher than what cocaine and heroin dealers have for gross margins. Real competition would be a good thing for everyone except Nvidia.

Reply View | 2 replies
- bigyabai 3 days ago
  
  80% isn't a ridiculous margin if nobody else is selling the same compute. Software margins famously go much higher, to as much as 95% or high-nines for cloud products. People pay the price hand over fist, because there simply isn't equivalent hardware to compete with. It's practically a steal for certain HPC customers that want the latest and greatest out of TSMC.
  I agree with both your comment and the parent comment - serious competition could spell the end for CUDA's dominance. But there will never be serious competition, CUDA has the head-start and their competitors threw in the towel with OpenCL. Khronos can't get Apple to sign onto a spec and they can't get AMD to change their architecture - open GPGPU compute is stuck in neutral while Nvidia is shifting into 6th gear. Reality is that Nvidia could charge cloud-level margins and get away with it, because Apple is the only other TSMC customer with equivalent leverage and they pretend the server market doesn't exist.
  
  Reply View | 0 replies
- brudgers 3 days ago
  
  Competing on price when the entrenched incumbent has 80% margins does not sell me on the idea.
  Or to put it another way, it would not be good for AMD.
  
  Reply View | 0 replies
joe_the_user 3 days ago

Finally, most of the call for AMD to develop a CUDA alternative is based on a desire for cheaper compute. That's not a good business venture to invest in against a dominate player because price sensitive customers are poor customers.
This is such a key point. Everyone wants cheaper and cheaper compute - I want cheaper and cheaper compute. But not large-ish company wants to simply facilitate cheapness - they would a significant return on their investment and just making a commodity is generally not what they want. Back in the days of the PC clone, the clone makers were relatively tiny and so didn't have to worry about just serving the commodity market.

Reply View | 1 reply
- brudgers 3 days ago
  
  The demand for clones was also diffuse…the potential market included SMB and consumers and the market was exponentially expanding. The compute market is scaling linearly among a relative few established players who buy in bulk and have long B2B relationships with nVidia.
  
  Reply View | 0 replies