Intel's Xe-HP GPU - The Big Boi

[The1Dickens]

We've heard about the Xe-LP, but us enthusiasts are far more interested in the opposite end of the spectrum. Can Intel give AMD and nVidia a run for their money?

Back in February, some information had been spotted for the 'High Performance' version of Intel's GPU, a 500-Watt variant. Some more information has come to light, including test drivers, about what the new architecture is capable of.
 

Quote

We have been hearing a lot about Intel's Xe GPUs but those are mostly the low power parts derived under the Xe-LP architecture. The Xe-HP 'High-Performance' GPUs were spotted by the tech leaker, Komachi (via HardwareLuxx), who found them hidden inside one of Intel's Open Source Platform documents retaining information about Iris Plus & UHD graphics. The updated adds a 'new systolic pipeline addition on EU (Execution Uni) from Gen 12 HP onwards' which are still early additions to the upcoming GPU architecture but it looks like Intel is now moving on from Xe-LP and is shifting gears to high-end graphics.

---

Quote

Previously released test drivers mentioned that Intel's Xe-LP architecture would power the DG1 or Discrete Graphics 1 GPUs while Xe-HP would power the DG2 or Discrete Graphics 2 GPUs. We don't know much about the DG2 tier of graphics cards except what was mentioned in the test drivers as they also mentioned what seemed to be execution units for each respective part. Following are the variants mentioned in the drivers:

• iDG1LPDEV = "Intel(R) UHD Graphics, Gen12 LP DG1" "gfx-driver-ci-master-2624"
• iDG2HP512 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"
• iDG2HP256 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"
• iDG2HP128 = "Intel(R) UHD Graphics, Gen12 HP DG2" "gfx-driver-ci-master-2624"

Three parts were mentioned with 128, 256 and 512 EUs (these could also be taken as bus widths but previous GPUs used this number to refer to the EU count, not bus width). Considering that DG1 with 96 EUs sits at around 2-3 TFLOPs, a 128 EU chip could end up around 4-5 TFLOPs with 256 EUs offering around 5-10 TFLOPs and 512 EUs offering around 10-15 TFLOPs of FP32 Compute output if clocks scale really well on the higher-end GPUs.

---

Quote

It looks like that Intel's Xe-HP Enthusiast and Workstation graphics cards would go up against AMD's RDNA2/CDNA2 and NVIDIA's Ampere GPUs which are expected to be announced later this year. Intel would be unveiling its Xe-LP powered stuff with Tiger Lake CPUs first and then move on to the discrete graphics cards offerings for consumer and workstation market and finally enter the exascale race by 2021.

WCCFTech 1 / 2

DigitalTrends (Feb)

 

I wonder if this is where all their 7nm process focus has been going, ensuring this thing is as powerful as possible, since the CPUs are doing pretty good (albeit stale) on whatever '+' level they are on now. I'm actually kind of excited for this. Also, to my knowledge, this is going to be the first PCIe 4.0 GPU. Not that we've really seen a need for that bandwidth, but it might be a significant indicator of how this card works. I am (barely) patiently waiting for the reviews of this, and wonder how it will process ray-tracing.

[Fnige]

3 minutes ago, The1Dickens said:

Intel's GPU, a 500-Watt variant

holy thats a lot of watts

 

What is it like compared to other cards?

[SenKa]

1 minute ago, SlimyPython said:

holy thats a lot of watts

 

What is it like compared to other cards?

TDP is not an exact science, and the numbers are not comparable. Buuut, thats double the power consumption of a 2080TI's TDP.

[The1Dickens]

Just now, SlimyPython said:

holy thats a lot of watts

 

What is it like compared to other cards?

A V100 is apparently 300 Watts. And the RTX Titan is 280 Watts.

[Fnige]

Just now, SenKa said:

TDP is not an exact science, and the numbers are not comparable. Buuut, thats double the power consumption of a 2080TI's TDP.

 

Just now, The1Dickens said:

A V100 is apparently 300 Watts. And the RTX Titan is 280 Watts.

 

So if Performance per TDP (I know its not a real measurement but just let me) is the same all round, intel has 2x the power which I doubt

 

Also its intel's TDP we're talking about here. I wonder how hot it'll run

[Arika]

6 minutes ago, SenKa said:

TDP is not an exact science, and the numbers are not comparable. Buuut, thats double the power consumption of a 2080TI's TDP.

it's also at 48Volts..christ

 

therefore not meant for consumers anyway

[5x5]

*insert double water chiller noises*

[Fnige]

46 minutes ago, Arika S said:

it's also at 48Volts..christ

 

therefore not meant for consumers anyway

[Master Disaster]

4 hours ago, The1Dickens said:

Also, to my knowledge, this is going to be the first PCIe 4.0 GPU. Not that we've really seen a need for that bandwidth, but it might be a significant indicator of how this card works.

rDNA uses PCIe 4.0 for the 2 high end cards (5700 & 5700XT).

[Letgomyleghoe]

Hmmm, hot processors, and now hot gpus, I see a trend

[porina]

The data centre parts are not necessarily representative of what we'll eventually see for mainstream consumption. 500W isn't a big deal, if it does the work you'd expect from 500W. The way Xe has been shown in the past is it looks more modular than we've seen in the GPU space. Almost chiplet like. So scaling should be relatively easy in this use case. It may be more of a challenge in consumer space later.

[Sauron]

7 hours ago, The1Dickens said:

WCCFTech

[mbntr]

500W bloody hell, it's a room heater, it's even worse than the GTX 480!

[mbntr]

4 hours ago, Master Disaster said:

rDNA uses PCIe 4.0 for the 2 high end cards (5700 & 5700XT).

also 5600 and 5500, but the 5500 is just PCIe x8, so it has the same bandwidth of PCIe x16 3.0

[RejZoR]

For as long as Intel brings competition, I'm interested. I just hope control panel won't be Intel's iGPU sadness... Where you can enable Anisotropic Filter. And that's about it since it doesn't even give you the multiplier options lol. I'm optimistic.