Tuesday, January 5th 2021

Intel DG2 Xe-HPG Features 512 Execution Units, 8 GB GDDR6

Intel's return to discrete gaming GPUs may have had a modest beginning with the Iris Xe MAX, but the company is looking to take a real stab at the gaming market. Driver code from the latest 100.9126 graphics driver, and OEM data-sheets pieced together by VideoCardz, reveal that its next attempt will be substantially bigger. Called "DG2," and based on the Xe-HPG graphics architecture, a derivative of Xe targeting gaming graphics, the new GPU allegedly features 512 Xe execution units. To put this number into perspective, the Iris Xe MAX features 96, as does the Iris Xe iGPU found in Intel's "Tiger Lake" mobile processors. The upcoming 11th Gen Core "Rocket Lake-S" is rumored to have a Xe-based iGPU with 48. Subject to comparable clock speeds, this alone amounts to a roughly 5x compute power uplift over DG1, 10x over the "Rocket Lake-S" iGPU. 512 EUs convert to 4,096 programmable shaders.

A leaked OEM data-sheet referencing the DG2 also mentions a rather contemporary video memory setup, with 8 GB of GDDR6 memory. While the Iris Xe MAX is built on Intel's homebrew 10 nm SuperFin node, Intel announced that its Xe-HPG chips will use third-party foundries. With these specs, Intel potentially has a GPU to target competitive e-sports gaming (where the money is). Sponsorship of major e-sports clans could help with the popularity of Intel Graphics. With enough beans on the pole, Intel could finally invest in scaling up the architecture to even higher client graphics market segments. As for availability, VideoCardz predicts a launch roughly coinciding with that of Intel's "Tiger Lake-H" mobile processor series, possibly slated for mid-2021.
Source: VideoCardz
Add your own comment

34 Comments on Intel DG2 Xe-HPG Features 512 Execution Units, 8 GB GDDR6

#26
dyonoctis
Patr!ckThis is just the beginning. I can't wait for a full Intel build with CPU + GPU + Optane SSD.
I was right to be cautious, the official documentation of Intel about GPU rendering on Blender only talks about OpenCL, it doesn't seem like their driver are going to run Optix/CUDA apps like it was native:
GPGPU: Blender (intel.com)
ZLUDA is a project that can make it possible, but Intel isn't involved:
ZLUDA: Drop-In Open-Source CUDA Support For Intel Xe / UHD Graphics - Phoronix

So it looks like it will once again be in the hands of the developers to use OneAPI instead of CUDA, and sucess where Open Cl failed. (Wich could be a good thing since OneAPI isn't exclusive to Intel.) The good news is that the CPU side of OneAPI is already used by a few popular render engines, so maybe there's hope.
Posted on Reply
#28
Borc
Chrispy_Tiger Lake's Xe IGP wasn't a disaster, and that was 25W comparable to Vega8.

If we take a Vega8 and multiply by 5.3 to emulate the shift from 96 to 512 EU, we get a "Vega42". That's not exactly bleeding edge, given that a Vega64 is starting to show its age, but 2/3rds of a Vega64 would probably put it in the ballpark of a GTX 1060 6GB or RX570.
Vega8 needs much higher GPU clocks to stay competitive against Xe LP. Even at 2100 Mhz it can lose in some against 1300 Mhz Xe LP. Also Xe LP is more bandwidth starved than Vega 8, the difference from DDR4 to LPDDR4 is quite big on Xe LP unlike on Vega8. The real gamer Xe will run on GDDR6 and won't be limited to DDR4/LPDDR4 and also it should clock a lot higher in a more power optimized environment, Xe LP is low clocked these days in comparison.
Posted on Reply
#29
lexluthermiester
BorcVega8 needs much higher GPU clocks to stay competitive against Xe LP.
This is true. It's somewhat amazing that this gets said but Intel is stepping up to the plate in the GPU game, and doing well. Kind of refreshing actually.
Posted on Reply
#30
londiste
BorcVega8 needs much higher GPU clocks to stay competitive against Xe LP. Even at 2100 Mhz it can lose in some against 1300 Mhz Xe LP. Also Xe LP is more bandwidth starved than Vega 8, the difference from DDR4 to LPDDR4 is quite big on Xe LP unlike on Vega8. The real gamer Xe will run on GDDR6 and won't be limited to DDR4/LPDDR4 and also it should clock a lot higher in a more power optimized environment, Xe LP is low clocked these days in comparison.
Vega 8 is 8CU - 512:32:16 (SP:TMU:ROP).
Big Xe is 96EU - 768:48:24.
Intel is using a 50% bigger iGPU configuration.
Posted on Reply
#31
Borc
The point is a simple multiplication to emulate the shift from 96 to 512 EU won't work unless he assumes a big dedicated GPU from Intel will stay on such a low GPU clock speed at 1300-1350 Mhz. It's a bit different if DG2 clocks at 2000 Mhz for example. Vega has been clock speed optimized for several generations, there is no real headroom anymore.
Posted on Reply
#32
lexluthermiester
BorcVega has been clock speed optimized for several generations, there is no real headroom anymore.
That's not true. But it's also not relevant anymore. AMD is moving away from Vega on it's iGPU packages. It is those advancements that Intel has to compete with.
Posted on Reply
#33
Borc
lexluthermiesterThat's not true. But it's also not relevant anymore. AMD is moving away from Vega on it's iGPU packages. It is those advancements that Intel has to compete with.
Why it isn't true? AMD uses Vega since Raven Ridge. Cezanne barely started, why it isn't relevant? Cezanne has to deal with ADL-P this year for some time.

Raven Ridge - max 1300 Mhz
Picasso - max 1400 Mhz
Renoir- max 1750 Mhz
Cezanne - max 2100 Mhz
Posted on Reply
Add your own comment
Dec 22nd, 2024 10:25 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts