Graphics Cores Past and Present; looking back at AMD's recent GPUs

Subject: Graphics Cards | September 5, 2018 - 05:50 PM |
Tagged: amd, GCN, R9 290X, r9 390x, R9 Fury X, RX VEGA 64

[H]ard|OCP have been examining the generational performance differences between GPUs, starting with NVIDIA and moving onto AMD.  In this review they compare Hawaii GCN 1.1, Fiji GCN 1.3 and Vega10 GCN 1.5 on a wide variety of games.  AMD is a more interesting case as they have made more frequent changes to their architecture, while at the same time tending towards mid-range performance as opposed to aiming for the high end of performance and pricing.  This has led to interesting results, with certain GCN versions offering more compelling upgrade paths than others.  Take a close look to see how AMD's GPUs have changed over the past five years.

View Full Size

"Wonder how much performance you are truly getting from GPU to GPU upgrade in games? We take GPUs from AMD and compare performance gained from 2013 to 2018. This is our AMD GPU Generational Performance Part 1 article focusing on the Radeon R9 290X, Radeon R9 390X, Radeon R9 Fury X, and Radeon RX Vega 64 in 14 games."

Here are some more Graphics Card articles from around the web:

Graphics Cards

Source: [H]ard|OCP

September 5, 2018 | 08:19 PM - Posted by VegaNowThatTheMinersBitTheDust (not verified)

Looks like the Vega 64 GPUs will have a nice period of FineWine(TM), probably a much longer period than any previous GCN based generations.

There is Still Vega's Explicit Primitive Shaders available for new games as that Primitive Shader IP is built into Vega's Hardware! But that Implicit Primitive Shader conversion path done through the drivers to convert Legacy Games Kernels over to automatically using Vega's Explicit Primitive Shader Hardware was canceled. AMD bit off way more than they could chew with that Implicit Primitive Shader to Explicit Primitive Shader conversion automation in their Vega drivers.

It looks like AMD would rather focus on Navi for gaming and let maybe the Games Developers decide if they will use that IP on Vega. And Navi will have all the IP that Vega has and then some new IP that's Navi only. I'll bet that AMD is so far along with Navi that AMD would rather Games Developers focus on that and maybe some of that Navi games development can be back ported to Vega based games. Vega 20 at 7nm is for the Professional Compute/AI market mostly with Navi again being mostly for mainstream only gaming.

September 5, 2018 | 09:23 PM - Posted by InfinityFabricAndBeyondForRed (not verified)

From Phoronix:

"It has been expected that Vega 20 would feature XGMI as a high-speed GPU interconnect alternative to PCI Express and that was firmed up today thanks to a new set of AMDGPU Linux driver patches.

XGMI is a peer-to-peer high-speed interconnect and is based on Infinity Fabric. XGMI is basically AMD's alternative to NVIDIA's NVLink for inter-connecting GPUs. Previously leaked slides indicated that XGMI would be supported on Vega 20 along with PCI Express 4.0. That appears accurate as becoming public today were a set of patches wiring in XGMI support to the AMDGPU Direct Rendering Manager driver and explicitly enabling the functionality for Vega 20. It's also expected AMD's upcoming Naples server CPU architecture will also support XGMI. " (1)

Note: I think he ment upcoming Rome Server CPU architecture as Naples is Zen and Rome is Zen2.

(1)

"Radeon Vega 20 Will Have XGMI - Linux Patches Posted For This High-Speed Interface"

https://phoronix.com/scan.php?page=news_item&px=AMDGPU-XGMI-Vega20-Patches

September 5, 2018 | 10:19 PM - Posted by InAdditionTaDiz (not verified)

Also from wikichip for the Zen CPU Micro-Arch:

"xGMI (Inter-Chip Global Memory Interconnect), " (1)

Which hyperlinks to the Infiinity fabtic wikichip entry so xGMI is Infinity fabric based and since this is listed in wikichip's Zen micro-arch listing(1) maybe xGMI(Infinity Fabric) will be available for Zen Epyc/Naples also, but that's probably only for server/workstations. So xGMI is just the infinity fabric used to wire up 2 Vega GPUs or a Zen/Epyc CPU to a Vega/Vega 20 GPU/s just like Nvidia's NVLink.

It's really all SERDES based with that wiring up everything on AMD's Zen and Vega SKUs and the various Data Protocol controllers(PCIe, IF, Other using the SERDES links).

Maybe at some point in time some Zen2/Navi consumer SKUs will make use of xGMI and Nvidia is sure bringing its NVLink to its consumer RTX GPU SKUs also. Hopefully with Both Nvidia(NVLink) and AMD(Infinity Fabric) between CPUs and dual/or more GPUs that will scale better than any driver based CF/SLI. Both NVLink and Infinity Fabric will enable cache coherency traffic between 2 GPUs just as its done for AMD's CPUs. Nvidia has no CPUs in the market for PC/Laptop gaming currently with only the ARM Core based Tegra portable gaming Nintendo Switch which is not using any newer Nvidia IP.

With AMD's Zen2 CPUs maybe using that xGMI(IF) IP between any Vega, or Navi, graphics that is going to behave differently if the Zen2 based consumer CPU SKUs And Vega/Navi GPUs can make use of xGMI instead of PCIe and Nvidia will have to use Zen/Zen2 to Nvidia GPU interfacing over PCIe because Nvidia does not support xGMI(Infinity Fabric).

(1)

"Zen - Microarchitectures - AMD"

https://en.wikichip.org/wiki/amd/microarchitectures/zen

September 6, 2018 | 11:28 AM - Posted by WhyMe (not verified)

I wish PCPer would implement some form of character limit in the comments section, the constant spamming with versions of war and peace is becoming tiresome.

Although seeing as 'that' guy has admitted he only does it to annoy and troll others he'd probably spend hours breaking his rants down into multiple posts and answering captcha all day long.

September 6, 2018 | 12:39 PM - Posted by HaHaTheMoronIsOffended (not verified)

F-ck off you whining moron, if you can't take it then just bugger off somewhere else! This is not twitter and those post are specifically designed to PISS YOU OFF. Just you keep up that whining about post limits. Whatever gets under your skin and get your goat is good and necessary!

September 7, 2018 | 02:24 AM - Posted by WhyMe (not verified)

It seems you're the only one getting upset as every time someone mentions your inane posts you launch into a full on expletive filled rant about how you're purposefully trolling people, how about you actually try making a comment that isn't just regurgitate information from other sites and insults?

September 21, 2018 | 05:48 PM - Posted by henrydevid

GPU is a great for all. So for that if you want to know more then just browse the site by which you will very helpful.Because it will very important topic for all. For more

https://uaewebsitedevelopment.com/apps-development/

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.