FP64 Performance and Separating Radeon VII from Radeon Instinct MI50

One of the interesting and amusing consequences of the Radeon VII launch is that for the first time in quite a while, AMD has needed to seriously think about how they’re going to differentiate their consumer products from their workstation/server products. While AMD has continued to offer workstation and server hardware via the Radeon Pro and Radeon Instinct series, the Vega 20 GPU is AMD’s first real server-grade GPU in far too long. So, while those products were largely differentiated by the software features added to their underlying consumer-grade GPUs, Radeon VII brings some new features that aren’t strictly necessary for consumers.

It may sound like a trivial matter – clearly AMD should just leave everything enabled – but as the company is trying to push into the higher margin server business, prosumer products like the Radeon VII are in fact a tricky proposition. AMD needs to lock away enough of the server functionality of the Vega 20 GPU that they aren’t selling the equivalent of a Radeon Instinct MI50 for a fraction of the price. On the other hand, it’s in their interest to expose some of these features in order to make the Radeon VII a valuable card in its own right (one that can justify a $699 price tag), and to give developers a taste of what AMD’s server hardware can do.

Case in point is the matter of FP64 performance. As we noted in our look at the Vega 20 GPU, Vega 20’s FP64 performance is very fast: it’s one-half the FP32 rate, or 6.9 TFLOPS. This is one of the premium features of Vega 20, and since Radeon VII was first announced back at CES, the company has been struggling a bit to decide how much of that performance to actually make available to the Radeon VII. At the time of its announcement, we were told that the Radeon VII would have unrestricted (1/2) FP64 performance, only to later be told that it would be 1/8. Now, with the actual launch of the card upon us, AMD has made their decision: they’ve split it down the middle and are doing a 1/4 rate.

Looking to clear things up, AMD put out a statement:

The Radeon VII graphics card was created for gamers and creators, enthusiasts and early adopters. Given the broader market Radeon VII is targeting, we were considering different levels of FP64 performance. We previously communicated that Radeon VII provides 0.88 TFLOPS (DP=1/16 SP). However based on customer interest and feedback we wanted to let you know that we have decided to increase double precision compute performance to 3.52 3.46 TFLOPS (DP=1/4SP).

If you looked at FP64 performance in your testing, you may have seen this performance increase as the VBIOS and press drivers we shared with reviewers were pre-release test drivers that had these values already set. In addition, we have updated other numbers to reflect the achievable peak frequency in calculating Radeon VII performance as noted in the [charts].

The end result is that while the Radeon VII won’t be as fast as the MI60/MI50 when it comes to FP64 compute, AMD is going to offer the next best thing, just one step down from those cards.

At 3.5 TLFLOPS of theoretical FP64 performance, the Radeon VII is in a league of its own for the price. There simply aren’t any other current-generation cards priced below $2000 that even attempt to address the matter. All of NVIDIA’s GeForce cards and all of AMD’s other Radeon cards straight-up lack the necessary hardware for fast FP64. The next closest competitor to the Radeon VII in this regard is NVIDIA’s Titan V, at more than 4x the price.

It’s admittedly a bit of a niche market, especially when so much of the broader industry focus is on AI and neural network performance. But there’s none the less going to be some very happy data scientists out there, especially among academics.

AMD Server Accelerator Specification Comparison
  Radeon VII Radeon Instinct
MI50
Radeon Instinct
MI25
FirePro S9170
Stream Processors 3840
(60 CUs)
3840
(60 CUs)
4096
(64 CUs)
2816
(44 CUs)
ROPs 64 64 64 64
Base Clock 1450MHz 1450MHz 1400MHz -
Boost Clock 1750MHz 1746MHz 1500MHz 930MHz
Memory Clock 2.0Gbps HBM2 2.0Gbps HBM2 1.89Gbps HBM2 5Gbps GDDR5
Memory Bus Width 4096-bit 4096-bit 2048-bit 512-bit
Half Precision 27.6 TFLOPS 26.8 TFLOPS 24.6 TFLOPS 5.2 TFLOPS
Single Precision 13.8 TFLOPS 13.4 TFLOPS 12.3 TFLOPS 5.2 TFLOPS
Double Precision 3.5 TFLOPS
(1/4 rate)
6.7 TFLOPS
(1/2 rate)
768 GFLOPS
(1/16 rate)
2.6 TFLOPS
(1/2 rate)
DL Performance ? 53.6 TFLOPS 12.3 TFLOPS 5.2 TFLOPS
VRAM 16GB 16GB 16GB 32GB
ECC No Yes (full-chip) Yes (DRAM) Yes (DRAM)
Bus Interface PCIe Gen 3 PCIe Gen 4 PCIe Gen 3 PCIe Gen 3
TDP 300W 300W 300W 275W
GPU Vega 20 Vega 20 Vega 10 Hawaii
Architecture Vega
(GCN 5)
Vega
(GCN 5)
Vega
(GCN 5)
GCN 2
Manufacturing Process TSMC 7nm TSMC 7nm GloFo 14nm TSMC 28nm
Launch Date 02/07/2019 09/2018 06/2017 07/2015
Launch Price (MSRP) $699 - - $3999

Speaking of AI, it should be noted that machine learning performance is another area where AMD is throttling the card. Unfortunately, more details aren’t available at this time. But given the unique needs of the ML market, I wouldn’t be surprised to find that INT8/INT4 performance is held back a bit on the Radeon VII. Or for that matter certain FP16 dot products.

Also on the chopping block is full-chip ECC support. Thanks to the innate functionality of HBM2, all Vega cards already have free ECC for their DRAM. However Vega 20 takes this one step further with ECC protection for its internal caches, and this is something that the Radeon VII doesn’t get access to.

Finally, Radeon VII also cuts back a bit on Vega 20’s off-chip I/O features. Though AMD hasn’t made a big deal of it up to now, Vega 20 is actually their first PCI-Express 4.0-capable GPU, and this functionality is enabled on the Radeon Instinct cards. However for Radeon VII, this isn’t being enabled, and the card is being limited to PCIe 3.0 speeds (so future Zen 2 buyers won’t quite have a PCIe 4.0 card to pair with their new CPU). Similarly, the external Infinity Fabric links for multi-GPU support have been disabled, so the Radeon VII will only be a solo act.

On the whole, there’s nothing very surprising about AMD’s choices here, especially given Radeon VII’s target market and target price. But these are notable exclusions that are going to matter to certain users. And if not to drive those users towards a Radeon Instinct, then they’re sure to drive those users towards the inevitable Vega 20-powered Radeon Pro.

Vega 20: Under The Hood Meet the AMD Radeon VII
POST A COMMENT

292 Comments

View All Comments

  • eva02langley - Thursday, February 7, 2019 - link

    I think you are the one in denial over this.

    This is a Radeon Instinct M150. This is a compute card that was never intended to be a gaming card. The biggest integration AMD had to do were drivers. Drivers will indeed be better in the next 3 months.
    Reply
  • yasamoka - Thursday, February 7, 2019 - link

    So please explain why AMD are designating this as a gaming card. I explained this in my previous post. Their lack of product differentiation is exhausting them and the apologetic acrobatics pulled by their die-hard fanboys is appallingly misleading. This is the same Vega architecture. Why isn't AMD releasing their cards with the drivers optimized beforehand? They have been doing this since the 7970. Remember that card getting beat by the GTX 680 just because they had unoptimized drivers? It took AMD almost a year to release drivers that thoroughly bested that series. CrossFire performance on these was also way ahead of SLi. It took them around another year to solve microstuttering. I had 2x 7970's back then. These delays need to stop, they're literally murdering AMD's product launches. Reply
  • Bp_968 - Thursday, February 7, 2019 - link

    Nvidia is also guilty of the same thing. The entire Turing lineup is a die designed and R&Ded for the Enterprise customers. The "flagship features" they keep raving about for Turing are slapped together ways to use all the ASIC cores on the dies designed for AI and content creation.

    Its why when you compare a 1080 to a 2070, or a 1080ti to a 2080 (the same price bracket) you get almost zero rasterization improvements. Its a huge and expensive die reused from the enterprise department because no one else has anything competitive in the same space.

    Nvidia is likely holding back their actual 12nm/7nm gamer design for 2020 out of concern for what Intel might have and possible concern over Navi. I also think Nvidia vastly underestimated how poorly the repackaged cards would sell. I expect Turing to be a very short generation with the next series being announced in late 2019 early 2020 (depending on what intel and AMD end up fielding).
    Reply
  • Alistair - Thursday, February 7, 2019 - link

    exactly! Reply
  • ToyzRUsKid - Friday, February 8, 2019 - link

    The updated NVENC encoder chip may become a major selling point for the RTX cards for streamers/content creators. I'm actually disappointed Nvidia is not emphasizing this feature more. Once OBS Studio releases their new build that will further increase NVENC encoding efficiency it will create an even more compelling argument to switch to NVENC.

    I have been testing the new encoder and it's rivaling and beating medium preset x264 at 1080p60 using 8k bitrate. Single pc streamers will see steam quality improvements along with massive cpu resource savings. I'm of the opinion the dark horse selling point of these cards will be the new NVENC encoder. It appears the Turning generation is more of an advancement for content creators than the average gamer. Ray tracing is superfluous at this point for sure.

    I'm ok with this. I still run a 1080Ti in my gaming rig and I'm comfortable waiting another generation. But the RTX 2070 in my streaming rig is delivering the best quality stream to date. That is comparing against x264 medium running on an i9 9900k@5GHz. This flies in the face of conventional wisdom and people with more credibility than me will need to help change the winds here. But this is my anecdotal experience.
    Reply
  • rahvin - Friday, February 8, 2019 - link

    If and I mean IF nvidia is holding back it's a purely financial move due to the huge overstock on GPU's caused by crypto currencies. Both AMD and nvidia massively underestimate how much demand crypto was creating. (IIRC AMD said during the earnings call now that crypto has dropped off that monthly GPU sales are less than half what they were) Supposedly there are more than 100K of nvidia cards (again the stuff I saw said that was somewhere between 3-6months normal gamer sales) sitting out there on store shelves rotting because of it, and it's so bad nvidia is having to take stock back from retailers that want the shelf space freed up.

    That's prime incentive to sit on the designs until the existing stock is used up. For both nvidia and AMD. Sure they might push out some high end high price product but they aren't doing anything in the middle of the market until that stock is cleared out.
    Reply
  • Korguz - Thursday, February 7, 2019 - link

    yasamoka...

    at the least.. this could force, what seems to be your saint nvida.. to drop the price of their cards.. as it stands before today.. ALL of their 20 series cards.. are out of the price range i would pay for a video card, or are pushing it/hard to justify the cost over my current 1060, as the 1070/80 were way out of my price range..
    Reply
  • D. Lister - Thursday, February 7, 2019 - link

    @Korguz

    The RVII performs 5%-6% below the 2080 (as per this review) and yet is priced the same. How is that going to force Nvidia to cut prices?
    Reply
  • Korguz - Friday, February 8, 2019 - link

    well.. where i am at least.. the radeon 7 starts at 949 ( preorder ), only 2 cards listed, Asus and xfx. the gtx 2080 ( which the radeon 7 is aimed at ) starts at 1130, almost 200 more... IMO.. 200 is not worth the premium for a 5-6% faster card.. the top of the line 2080 is the GeForce RTX 2080 SEA HAWK X which is priced at $1275... for my cash... id be looking at the radeon 7.. and saving 200+ bucks to use some where else in my comp... Reply
  • TheinsanegamerN - Thursday, February 7, 2019 - link

    The Instinct M150 is a VEGA class card, the architecture is incredibly similar to Vega 56/64. There are not massive gains to be made here.

    If there are, then AMD must be completely incompetent at driver management.
    Reply

Log in

Don't have an account? Sign up now