Wednesday, April 12th 2023
AMD Plays the VRAM Card Against NVIDIA
In a blog post, AMD has pulled the VRAM card against NVIDIA, telling potential graphics card buyers that they should consider AMD over NVIDIA, because current and future games will require more VRAM, especially at higher resolution. There's no secret that there has been something of a consensus from at least some of the PC gaming crowd that NVIDIA is being too stingy when it comes to VRAM on its graphics cards and AMD is clearly trying to cash in on that sentiment with its latest blog post. AMD is showing the VRAM usage in games such as Resident Evil 4—with and without ray tracing at that—The Last of US Part I and Hogwarts Legacy, all games that use over 11 GB of VRAM or more.
AMD does have a point here, but as the company has as yet to launch anything below the Radeon RX 7900 XT in the 7000-series, AMD is mostly comparing its 6000-series of cards with NVIDIA's 3000-series of cards, most of which are getting hard to purchase and potentially less interesting for those looking to upgrade their system. That said, AMD also compares its two 7000-series cards to the NVIDIA RTX 4070 Ti and the RTX 4080, claiming up to a 27 percent lead over NVIDIA in performance. Based on TPU's own tests of some of these games, albeit most likely using different test scenarios, the figures provided by AMD don't seem to reflect real world performance. It's also surprising to see AMD claims its RX 7900 XTX beats NVIDIA's RTX 4080 in ray tracing performance in Resident Evil 4 by 23 percent, where our own tests shows NVIDIA in front by a small margin. Make what you want of this, but one thing is fairly certain and that is that future games will require more VRAM, but most likely the need for a powerful GPU isn't going to go away.
Source:
AMD
AMD does have a point here, but as the company has as yet to launch anything below the Radeon RX 7900 XT in the 7000-series, AMD is mostly comparing its 6000-series of cards with NVIDIA's 3000-series of cards, most of which are getting hard to purchase and potentially less interesting for those looking to upgrade their system. That said, AMD also compares its two 7000-series cards to the NVIDIA RTX 4070 Ti and the RTX 4080, claiming up to a 27 percent lead over NVIDIA in performance. Based on TPU's own tests of some of these games, albeit most likely using different test scenarios, the figures provided by AMD don't seem to reflect real world performance. It's also surprising to see AMD claims its RX 7900 XTX beats NVIDIA's RTX 4080 in ray tracing performance in Resident Evil 4 by 23 percent, where our own tests shows NVIDIA in front by a small margin. Make what you want of this, but one thing is fairly certain and that is that future games will require more VRAM, but most likely the need for a powerful GPU isn't going to go away.
218 Comments on AMD Plays the VRAM Card Against NVIDIA
Some of the RAM fudgery was due to shortages and the pandemic too, but i'd be lying if i said it didn't also annoy me - plus AMD PCB quality has only just improved and their silicon use is generally of a lower quality/higher yield type. Despite all the amazing PR and Advertising AMD do, they still haven't actually "won" anything yet, even if they had a few things first (that again, only matter now, when those machines are no longer fast enough to take full advantage of it) etc.. I fell for the "native quad core" and "integrated memory controller" crap too almost, i was a total AMD sucker.. Haha! Pretty sure nVidia copied AMD with the whole amount of RAM on the 3060 vs the Ti though.. 12GB on a 3060 and 2060 why? And indeed 4GB of their new HBM invention on the Fury helped kill it...that and it was kind of shit but...that was their best early chance at actually countering nVidia if they had added more RAM or used GDDR funnily enough! But thanks to HBM, you can't really even mod them easily to change the RAM! They certainly have plenty of income sources that aren't PCs to keep competitive though, monopoly on 2 of the 3 popular consoles etc.. AMD have become rich by being devious and doing deals, not by being the best at what they do. Not like nobody does that but AMD, they lie to the customers far more than most. How's your "eight core" FX?
RDNA 2 on the other hand, is a hands down excellent line of products, on par with Nvidia in rasterization, stability, updates, etc, and is cheaper, too. I also like the AMD driver way more than Nvidia's Windows 95 style retro control panel with the way too convoluted 3D Settings menu. The only thing that's lacking is RT performance, but that's not great on any current card except for maybe the 4090, anyway.
IE we helped Nvidia prep for AI.
But don't worry Nvidia can give you back free frames inserted between the two in your Vram buffer that will make it go vroom, at least until you 180 faster than casual.
I never did like v-sync or triple buffering all added too much lag and still do.
8 GB VRAM-equipped PC GPU card can't match PS5's random access VRAM range and scope. VRAM problem is not an issue for NVIDIA owners with 16 GB VRAM or greater.
Read
www.forbes.com/sites/davealtavilla/2022/10/24/nvidia-geforce-rtx-4090-sell-through-indicators-highlight-a-thriving-pc-gaming-market/?sh=362fe7f343a0
www.tomshardware.com/news/nvidia-maintains-lead-as-sales-of-graphics-cards-hit-all-time-low-in-2022-jpr
Dr. Jon Peddie, president of JPR. "Some products like Nvidia’s RTX 4090 did exceptionally well despite its high price, so almost everything we thought we knew about economics and market behavior seemed to be turned on its head in Q4.”
NVIDIA's VRAM tactics worked since RTX 4090 has 24 GB of VRAM.
Anyway, i think the other reason is...ready for a slight conspiracy? Crypto... AMD gave more RAM on their Radeon cards after the 290/X series, unlike nVidia, because.. Even though the card wasn't fast enough to actually use 8GB VRAM and run above 1080p or 4K etc, including the RX 4/5 series here as they're basically all the same with some small revisions.. But they are more useful for crypto mining! Only reason i couldn't use my old GTX 970 for mining is it didn't have enough VRAM, not that i'd bother mining..but it certainly made their cards more useful for the...wrong reasons. My 290X was of course faster than my 970 that replaced it, but the 970 is still working well as is the SSD i got at the same time, whereas the 290X died after a couple of months and arrived with damaged fans - it was an MSi card which says it all really.
Heavy VRAM usage is useful for the game's fundamental artwork.
PS4 / PS4 Pro era has 4 GB to 5.5GB VRAM game design hence 6 GB to 8 GB VRAM is good enough for PCs until PS5 / XSX reached sizable numbers e.g. 58 million.
Xbox Series X was designed with a 10 GB fast memory range as its primary VRAM range. Xbox Series X allocates 13.5 GB of memory to Games
Consoles don't have PC duplicated data problems.
READ www.techpowerup.com/306713/directx-12-api-new-feature-set-introduces-gpu-upload-heaps-enables-simultaneous-access-to-vram-for-cpu-and-gpu?cp=2
Microsoft has implemented two new features into its DirectX 12 API - GPU Upload Heaps and Non-Normalized sampling have been added via the latest Agility SDK 1.710.0 preview, and the former looks to be the more intriguing of the pair. The SDK preview is only accessible to developers at the present time, since its official introduction on Friday 31 March. Support has also been initiated via the latest graphics drivers issued by NVIDIA, Intel, and AMD. The Microsoft team has this to say about the preview version of GPU upload heaps feature in DirectX 12: "Historically a GPU's VRAM was inaccessible to the CPU, forcing programs to have to copy large amounts of data to the GPU via the PCI bus. Most modern GPUs have introduced VRAM resizable base address register (BAR) enabling Windows to manage the GPU VRAM in WDDM 2.0 or later."
They continue to describe how the update allows the CPU to gain access to the pool of VRAM on the connected graphics card: "With the VRAM being managed by Windows, D3D now exposes the heap memory access directly to the CPU! This allows both the CPU and GPU to directly access the memory simultaneously, removing the need to copy data from the CPU to the GPU increasing performance in certain scenarios." This GPU optimization could offer many benefits in the context of computer games, since memory requirements continue to grow in line with an increase in visual sophistication and complexity.
A shared pool of memory between the CPU and GPU will eliminate the need to keep duplicates of the game scenario data in both system memory and graphics card VRAM, therefore resulting in a reduced data stream between the two locations. Modern graphics cards have tended to feature very fast on-board memory standards (GDDR6) in contrast to main system memory (DDR5 at best). In theory the CPU could benefit greatly from exclusive access to a pool of ultra quick VRAM, perhaps giving an early preview of a time when DDR6 becomes the daily standard in main system memory.
---
Your argument is based on ignorance.
AMD's Fusion is coming for gaming PC.
And i couldn't give a fuck about AMD. Go away Lisa. Bad enough she fucked up chip designs in the 80s!
You're ignorant if you think i got a 3090 just for gaming! :roll:
XSX and PS5 target lower price and near-idiot-proof box gamer market. Digital edition PS5 (with 16 GB GDDR6-14000 and 512 DDR4) has a $399 asking price and RTX 3090 wouldn't match that price. RTX 3090 is not functional without a host x86 PC.
I didn't assume RTX 3090 is just used for games, that's your assumption.
"8 GB VRAM" wouldn't be able to match the console's 10 to 12 GB VRAM random access pattern.
RX 480 / RX 580 was under Raja Koduri's administration. For raster and before Raja Koduri, R9 290X reached 64 ROPS. Under Raja "Mr TFLOPS" Koduri, the raster was stalled at 64 ROPs into Vega 64, Vega II, and RX 5700 XT generations.
Lisa Su wasn't responsible for AMD's failed Bulldozer.
Lisa Su's team was responsible for CELL which is a flawed design along with NVIDIA's flawed 32-bit compute incompetent GeForce 7-based RSX. Don't assume I'm not aware of the PowerPC debacle, Hint: Amiga PowerPC.
Not reading much of the other crap you're typing to be honest, my GPU has GDDR6X so i don't care, but.. At least my PS4 Pro Dev/Test kit can be made more useful with Linux.. Just a shame the GPU driver only really works properly on the Slim i have due to different GPU archs, of course.
Bulldozer will go down in history as the biggest waste of Silicon ever made. And your asking price is bollocks, PS5s were £850 for a long time as was XSX. lol.
Also, anyone have any idea how much my PS4 Pro is worth? :roll:I only paid £1139 for my 3090 mate as i sold my 1080Ti with custom cooling for £500 ;P
Oh yeah, that was it, CELL! God that was a piece of shit but my PS3s are only useful due to being jailbroken.. PS5 has sat in its box for years waiting for that one glorious day we can exploit it! I was patient with my PS4s and it paid off!
Did you forget about Lisa Su's earlier work on something like the Atari chips or some other early computer/console systems? Have to google which now but she fucked it. :P Pretty sure it was Jay Miner that did the Atari 7800 chips actually.. No wonder i like mine! Can't say i like Amiga any more though, such poor system design for the time really, they were bound to fail and good god the OS and hardware is often about as stable as an Italian taxi driver that's got stuck behind two old priests in a Skoda!
en.wikipedia.org/wiki/Jay_Miner
RIP.. <3
The UK is a small country and I don't care i.e. that's your problem.
PS4 / PS4 Pro with modified desktop Linux is slow due to netbook-class Jaguar CPUs.
www.whathifi.com/advice/ps5-vs-ps5-digital-edition-which-should-you-buy
Article date: November 24, 2022
Standard PlayStation 5 debuted at £449 / $499 / AU $749
November 24, 2022: £479.99 / $499 / AU $799.95.
The disc-less PS5 Digital Edition, Sony's August 2022 price rises have nudged that up to £389.99 / $399.99 / AU $649.95.
Paying more than £479.99 is not the official price. You're the real bollocks.Lisa Su wasn't at Atari and She wasn't old enough for developing 1980s game consoles. I paid very little attention to Atari 7800 and its shovelware games and the best 8-bit game console was Nintendo Entertainment System.
Lisa Su's summer jobs at Analog Devices.
Lisa Su obtained her master's degree from MIT in 1991. From 1990 to 1994, she studied for her Ph.D. under MIT advisor Dimitri Antoniadis.
In June 1994, Su became a member of the technical staff at Texas Instruments.
Atari's downfall is during Jack Tramiel's Commodore MK2 aka Atari Corporation.
CEO Jack Tramiel's cost-cutting was the downfall of Commodore Semiconductor Group(CSG)'s MOS Technology 65xx CPU family design that caused Acorn to develop the ARM CPU. LOL.
The main reason for switching from CSG/MOS 65xx CPU family to Motorola's 68000 CPU family is due to CEO Jack Tramiel's cost-cutting the CPU R&D. CSG/MOS 65xx couldn't keep up with Intel's X86 evolution.
Commodore-International Limited and CEO Jack Tramiel (and his Commodore Mk2 Atari Corporation) are the major factors, NOT Lisa Su!
Lisa Su is judged when She is the CEO of a company!
Your narrative is a load of crap.Read "Race for a New Game Machine: Creating the Chips Inside the Xbox 360 and the PlayStation 3" book which led design for IBM's PPE and SPE.
David Shippy was one of the lead architects for the POWER2*, G3 PowerPC, and POWER4* processor designs. He is the chief architect for the power processing unit for the Cell processor. www.goodreads.com/book/show/22121796-race-for-a-new-game-machine
Lisa Su's role is to represent IBM when they interface with Sony and Toshiba.
For a raster graphics workload, STI's CELL's SPU wasn't designed like ATI's Xenos GpGPU. IBM PPE's effective GFLOPS claims are rubbish i.e. they are worst than Jaguar CPUs. IBM lacks modern GPU design experience when compared to AMD(ATI) or NVIDIA or VIA S3.
From forum.beyond3d.com/threads/how-much-work-must-the-spus-do-to-compensate-for-the-rsxs-lack-of-power.48995/page-6#post-1460125
Against NVIDIA GeForce 7 and PS3's RSX GPU design flaws
------------------------
Unmasking NVIDIA's "The Way its Meant to be Played" during NVIDIA's GeForce 7 series.
"I could go on for pages listing the types of things the spu's are used for to make up for the machines aging gpu, which may be 7 series NVidia but that's basically a tweaked 6 series NVidia for the most part. But I'll just type a few off the top of my head:"NVIDIA's GeForce 8 (CUDA) series is a large improvement.
AmigaOS wasn't designed with MMU since 68000 did NOT have an MMU. LOL "Old school" Unix vendors with 68000 CPU-based workstations used a custom MMU until 68010's slow 68451 and 68020's 68851 add-on MMUs. 68030 was released in 1987 with a built-in MMU. Motorola wasn't taking Unix seriously and 68030 was late to the party. Many "Old school" Unix vendors started their RISC CPU R&D due to Motorola's inferior R&D roadmap.
Commodore's toy mindset has MMU less Amiga 1200's 68EC020 CPU baseline and doubled down with Amiga 4000 with MMU-less 68EC030. Motorola shouldn't offer MMU-less 68EC030.
Microsoft Xenix and Intel 80286 (with built-in MMU since 1982) dominated UNIX shipments in the 1980s. Intel 80386 has built-in MMU for Xenix, Linux, and Windows NT. Linux originated from an 80386-based PC. Unlike Motorola, Intel is consistent with the built-in MMUs with 286 and 386 to the current date.Intel has E-Cores... Look in the mirror. hypocrite.
Bulldozer Chief Architect removed from AMD, techreport.com/forums/viewtopic.php?t=85582 (Date: Dec 24, 2012)
Mike Butler, Chief Architect of the Bulldozer architecture, apparently doesn't work for AMD anymore.
Mike Butler is currently an Architect at Samsung. www.linkedin.com/in/mike-butler-049bb633
Meanwhile, in the real world and beyond HU donuts...
AMD did not withdraw the donuts from the stall if they did not smell suspicious. I know that they are vulnerable with this subject and act through intermediaries, such as HU.I had FX 8300 and I wanted to brag about multithreading. Eight cores, at that time, you could only find Intel processors 10 times more expensive.
Disaster!
The second disaster came when the little Pentium G4560 destroyed this FX in games with GTX 960.
It certainly doesn't compare to E-cores, "8 cores" being one of AMD's big lies, much bigger than the one now, with vRAM.
E-cores from 14700K outperform 3700X in multicore tests. They are not really small. These efficient cores made AMD supporters avoid the subject of renderings, encodings and multitasking.
My 13500 fights with 7700/X and not with 7600/X thanks to these E-cores.
E-cores independently beating a 3700X.... thats extremely impressive! Absolutely, if i were buying a system today with multithreading performance in mind intels 13th gen would be right up there in the list of options. Either that or the 7950X with a cheaper cooler + forward Gen upgrade support on AM5. Can't go wrong with either option.
Are you still defending 8GB VRAM for everyone? Honestly i respect this type of obstinate confirmation to live by the 8GB sword. If its perfectly in line with your personal gaming preferences and performance goals (and going forward), that's an amazing place to be. Some of us demand more and end up paying the penalty of higher premiums. We need more Gica's! 8-G-B for life!
And to post a video that shows that a $500 GPU is sometimes faster than a $400 GPU and sometimes not. Not really a good argument for the $500 GPU.
You seem confused.
12GB cards are aging better than 8GB cards. No brainer.