Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

arstechnica.com

Post reply on HN
Print view View original

61–70 of 218 posts

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#61

Post by mey »

pier25 wrote: post #54
Exmoor wrote: post #11

As TFA rightly points out, unless something drastically changes in the next ~6mo, Intel is going to launch into the most favorable market situation we've seen in our lifetimes. Previously, the expectation is that they needed to introduce something that was competitive with the top end cards from nVidia and AMD. With basically all GPU's out of stock currently they really just need to introduce something competitive wi…

Exactly. There are plenty of people that just want to upgrade an old GPU and anything modern would be a massive improvement. I'm still rocking a 1070 for 1080p/60 gaming and would love to jump to 4K/60 gaming but just can't convince myself to buy a new GPU at current prices.

I refuse to engage with the current GPU pricing insanity, so my 5900x is currently paired with a 960 GTX. When Intel enters the market it will be another factor in driving pricing back down, so might play Cyberpunk in 2022...

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#62

Post by judge2020 »

cwizou wrote: post #21

They still are not saying with which part of that lineup they want to compete with, which is a good thing. I still remember Pat Gelsinger telling us over and over that Larrabee would compete with the high end of the GeForce/Radeon offering back in the days, including when it was painfully obvious to everyone that it definitely would not. https://en.wikipedia.org/wiki/Larrabee_(microarchitecture)

Well there's already the DG1 which seems to compete with the low-end. https://www.youtube.com/watch?v=HSseaknEv9Q

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#63

Post by wmf »

BuckRogers wrote: post #57
opencl wrote: post #23

Earlier quoted context omitted.

Intel is not even manufacturing these, they are TSMC 7nm so they are competing for the same fab capacity that everyone else is using.

This is a problem for AMD especially, but also Nvidia. Not so much for Intel. They're just budging in line with their superior firepower. Intel even bought out first dibs on TSMC 3nm out from under Apple. I'll be interested to see the market's reaction to this once everyone realizes that Intel is hitting AMD where it hurts and sees the inevitable outcome. This is one of the smartest moves by Intel, make their own stu…

This is a problem for AMD especially

Probably not. AMD has had their N7/N6 orders in for years.

They're just budging in line with their superior firepower. Intel even bought out first dibs on TSMC 3nm out from under Apple.

There's no evidence this is happening and people with TSMC experience say it's not happening.

Nvidia and especially AMD took a risk not being in the fabrication business

Yes, and it paid off dramatically. If AMD stayed with their in-house fabs (now GloFo) they'd probably be dead on 14nm now.

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#64

Post by paulmd »

kllrnohj wrote: post #52
paulmd wrote: post #44

Earlier quoted context omitted.

if NVIDIA cards were priced as ridiculously as AMD cards they'd be sitting on store shelves too

Nvidia doesn't price any cards other than the founder's editions which you'll notice they both drastically cut down on availability for and also didn't do at all for the "price sensitive" mid-range tier. Nvidia's pricing as a result is completely fake. Like the claimed "$330 3060" in fact starts at $400 and rapidly goes up from there, with MSRP's on 3060's as high as $560.

I didn't say NVIDIA did directly price cards? Doesn't sound like you are doing a very good job of following the HN rule - always give the most gracious possible reading of a comment. Nothing I said directly implied that they did, you just wanted to pick a bone. It's really quite rude to put words in people's mouths, and that's why we have this rule.

But a 6900XT is available for $3100 at my local store... and the 3090 is $2100. Between the two it's not hard to see why the NVIDIA cards are selling and the AMD cards are sitting on the shelves, the AMD cards are 50% more expensive for the same performance.

As for why that is - which is the point I think you wanted to address, and decided to try and impute into my comment - who knows. Price are "sticky" (retailers don't want to mark down prices and take a loss) and AMD moves fewer cards in general. Maybe that means that prices are "stickier for longer" with AMD. Or maybe it's another thing like Vega where AMD set the MSRP so low that partners can't actually build and sell a card for a profit at competitive prices. But in general, regardless of why - the prices for AMD cards are generally higher, and when they go down the AMD cards sell out too. The inventory that is available is available because it's overpriced.

(and for both brands, the pre-tariff MSRPs are essentially a fiction at this point apart from the reference cards and will probably never be met again.)

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#65

Post by hobofan »

rowanG077 wrote: post #20
byefruit wrote: post #9

I really hope this breaks Nvidia's stranglehold on deep learning. Some competition would hopefully bring down prices at the compute high-end. AMD don't seem to even be trying on the software-side at the moment. ROCm is a mess.

I think this situation can only be fixed by moving up into languages that compile to vendor specific GPU languages. Just treat CUDA, openCL, vulkan compute, metal compute(??) etc. as the assembly of graphics cards.

Barely anyone is writing CUDA directly these days. Just add support in PyTorch and Tensorflow and you've covered probably 90% of the deep learning market.

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#66

Post by wmf »

dragontamer wrote: post #35
byefruit wrote: post #29

Earlier quoted context omitted.

Unfortunately that support is via ROCm, which doesn't support the last three generations (!) of AMD hardware: https://github.com/ROCm/ROCm.github.io/blob/master/hardware....

ROCm supports Vega, Vega 7nm, and CDNA just fine. The issue is that AMD has split their compute into two categories: * RDNA -- consumer cards. A new ISA with new compilers / everything. I don't think its reasonable to expect AMD's compilers to work on RDNA, when such large changes have been made to the architecture. (32-wide instead of 64-wide. 1024 registers. Etc. etc.) * CDNA -- based off of Vega's ISA. Despite bei…

Hopefully CDNA2 will be similar enough to RDNA2/3 that the same software stack will work with both.

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#67

Post by BuckRogers »

wmf wrote: post #63
BuckRogers wrote: post #57

Earlier quoted context omitted.

This is a problem for AMD especially, but also Nvidia. Not so much for Intel. They're just budging in line with their superior firepower. Intel even bought out first dibs on TSMC 3nm out from under Apple. I'll be interested to see the market's reaction to this once everyone realizes that Intel is hitting AMD where it hurts and sees the inevitable outcome. This is one of the smartest moves by Intel, make their own stu…

This is a problem for AMD especially Probably not. AMD has had their N7/N6 orders in for years. They're just budging in line with their superior firepower. Intel even bought out first dibs on TSMC 3nm out from under Apple. There's no evidence this is happening and people with TSMC experience say it's not happening. Nvidia and especially AMD took a risk not being in the fabrication business Yes, and it paid off dramat…

Do you have sources for any of your claims? Other than going fabless being a fantastic way to cut costs and management challenges, but increase longterm supply line risk, none of that is anything that I've heard. Here are sources for my claims.

AMD on TSMC 3nm for Zen5. Will be squeezed by Intel and Apple- https://videocardz.com/newz/amd-3nm-zen5-apus-codenamed-stri...

Intel consuming a good portion of TSMC 3nm- https://www.msn.com/en-us/news/technology/intel-locks-down-a...

I see zero upside with these developments for AMD, and to a lesser degree, Nvidia, who are better diversified with Samsung and also rumored to be in talks with fabricating at Intel as well.

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#68

Post by dragontamer »

wmf wrote: post #66
dragontamer wrote: post #35

Earlier quoted context omitted.

ROCm supports Vega, Vega 7nm, and CDNA just fine. The issue is that AMD has split their compute into two categories: * RDNA -- consumer cards. A new ISA with new compilers / everything. I don't think its reasonable to expect AMD's compilers to work on RDNA, when such large changes have been made to the architecture. (32-wide instead of 64-wide. 1024 registers. Etc. etc.) * CDNA -- based off of Vega's ISA. Despite bei…

Hopefully CDNA2 will be similar enough to RDNA2/3 that the same software stack will work with both.

I assume the opposite is going on.

Hopefully the RDNA3 software stack is good enough that AMD decides that CDNA2 (or CDNA-3) can be based off of the RDNA-instruction set.

AMD doesn't want to piss off its $100 million+ customers with a crappy software stack.

---------

BTW: AMD is reporting that parts of ROCm 4.3 are working with the 6900 XT GPU (suggesting that RDNA code generation is beginning to work). I know that ROCm 4.0+ has made a lot of github checkins that suggest that AMD is now actively working on the RDNA-code generation. Its not officially written into the ROCm documentation yet, its mostly the discussions with ROCm github issues that are noting these changes.

Its not official support and its literally years late. But its clear what AMD's current strategy is.

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#69

Post by davidjytang »

ayngg wrote: post #59
Exmoor wrote: post #11

As TFA rightly points out, unless something drastically changes in the next ~6mo, Intel is going to launch into the most favorable market situation we've seen in our lifetimes. Previously, the expectation is that they needed to introduce something that was competitive with the top end cards from nVidia and AMD. With basically all GPU's out of stock currently they really just need to introduce something competitive wi…

I thought they are using TSMC for their gpu, which means they will be part of the same bottleneck that is affecting everyone else.

I believe nvidia doesn't use TSMC or not only use TSMC.

Re: Intel’s Arc GPUs will compete with GeForce and Radeon in early 2022

#70

Post by hprotagonist »

hobofan wrote: post #65
rowanG077 wrote: post #20

Earlier quoted context omitted.

I think this situation can only be fixed by moving up into languages that compile to vendor specific GPU languages. Just treat CUDA, openCL, vulkan compute, metal compute(??) etc. as the assembly of graphics cards.

Barely anyone is writing CUDA directly these days. Just add support in PyTorch and Tensorflow and you've covered probably 90% of the deep learning market.

and ONNX.
Post reply on HN