• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

AMD 24.3.1 Drivers Unlock RX 7900 GRE Memory OC Limits, Additional Performance Boost Tested

Joined
Apr 18, 2019
Messages
2,155 (1.15/day)
Location
Olympia, WA
System Name Sleepy Painter
Processor AMD Ryzen 5 3600
Motherboard Asus TuF Gaming X570-PLUS/WIFI
Cooling FSP Windale 6 - Passive
Memory 2x16GB F4-3600C16-16GVKC @ 16-19-21-36-58-1T
Video Card(s) MSI RX580 8GB
Storage 2x Samsung PM963 960GB nVME RAID0, Crucial BX500 1TB SATA, WD Blue 3D 2TB SATA
Display(s) Microboard 32" Curved 1080P 144hz VA w/ Freesync
Case NZXT Gamma Classic Black
Audio Device(s) Asus Xonar D1
Power Supply Rosewill 1KW on 240V@60hz
Mouse Logitech MX518 Legend
Keyboard Red Dragon K552
Software Windows 10 Enterprise 2019 LTSC 1809 17763.1757
Well, this is weird. Today I wanted to play some more Horizon Forbidden West and after the pushing the card to its limits yesterday I am suddently getting intense artifacting and crashes all the way down to 2550 MHz memory clock. Whereas I played at 2664 MHz for almost 5 hours a few days ago, with no artifacts at all...

Do you guys think I might have degraded the chips through that? I didn't monitor memory temps but I made sure to set the fans on the card to 60% which was already plenty seeing as my core maxed out at 50C.

A Timespy run at 950mV/2530MHz mem just made my whole system lock up.
Don't look @ 'core'. Look @ hotspot (and memory) temps. I've seen 90-100c hotspot with 50-70c 'core' temps before. Lower vcore a lil, and the delta shrinks.

It 'shouldn't' have let you cook it, but you may have. I'd test w/ a clean driver re-install, or another PC.
(Also, I have had issues w/ loose PCIe power causing freezing, no artifacts tho)

Could be 'cold solder joints' or thermal warping too, if the card heatsoaked too much.
What you describe is 'similar' to a lot of people's VIIs when they died (from warping and cold joints, etc).

Thankfully, the GRE's not been out long enough to be outside of warranty.
While you were 'OCing' (which is typ. outside of warranty), the card has hard thermal limits to prevent damage. Potentially, those limits did not 'save' your card, and (personally) I'd not feel bad RMAing it.
 
Joined
May 3, 2018
Messages
2,400 (1.08/day)
This is interesting. I would normally say just get the 7900XT but I'm seeing a $250 price difference on the Powercolour Hellhound versions of GRE and XT. Let's say worst case 2600MHz OC is it enough of a jump over the 6800XT to bother, or go the XT route and save for an extra few weeks. I would rather the XTX but the jump in prices is too much IMO.

Nvidia's pricing is insane with 4070 Ti Super way dearer than 7900XT and 4080 Super starting at $1900 way dearer than XTX.
 
Joined
Mar 24, 2024
Messages
86 (1.21/day)
Location
Europe

insane some games almost Rx 7900XT performance . We got a new value King here.
 

FordGT90Concept

"I go fast!1!11!1!"
Joined
Oct 13, 2008
Messages
26,259 (4.60/day)
Location
IA, USA
System Name BY-2021
Processor AMD Ryzen 7 5800X (65w eco profile)
Motherboard MSI B550 Gaming Plus
Cooling Scythe Mugen (rev 5)
Memory 2 x Kingston HyperX DDR4-3200 32 GiB
Video Card(s) AMD Radeon RX 7900 XT
Storage Samsung 980 Pro, Seagate Exos X20 TB 7200 RPM
Display(s) Nixeus NX-EDG274K (3840x2160@144 DP) + Samsung SyncMaster 906BW (1440x900@60 HDMI-DVI)
Case Coolermaster HAF 932 w/ USB 3.0 5.25" bay + USB 3.2 (A+C) 3.5" bay
Audio Device(s) Realtek ALC1150, Micca OriGen+
Power Supply Enermax Platimax 850w
Mouse Nixeus REVEL-X
Keyboard Tesoro Excalibur
Software Windows 10 Home 64-bit
Benchmark Scores Faster than the tortoise; slower than the hare.
This is however fascinating, since I guess I just hit an ASRock which doesn't want to go over 2400MHz, which would go really closed to the reported instability issues around 2500MHz for some parts.

Which would mean, that some of the worldwide stock is repurposed 7900GRE for China which didn't sell (like mine). It looks like a very confusing case, because I feel that should be refundable... It's not even silicon lottery, it's a hidden-revision-that-is-not-in-specs at all. Either purposefully locked based on BW or binned based on max stable memory controller speed.
The GDDR modules installed on the card are up to the AIB partner's discretion so long as they meet or exceed reference spec. In this case, PowerColor (see OP) has much more capable Samsung chips installed compared your ASRock model. In short, ASRock skimped on the GDDR and PowerColor did not (likely used the same modules the 7900 XT and XTX have). All AMD did was enable overclocking of the VRAM so you can unlock whatever capability the GDDR modules have from the factory that were previously limited by the VBIOS.

That said, people shouldn't go into GRE thinking every card's VRAM is going to be able to overclock up to 7900 XT spec. AIBs like changing cards up based on GDDR chip availability/pricing without public notice. Stock clocks are all that is guaranteed/warrantied.


It's still not clear to me why AMD is doing this unless consumer overclocks are exempt from the 600 GB/s regulation.


Lots of really fascinating info/takes is this thread, thanks all for your contributions (and for helping inform people/make the whole ecosystem better).

I, for one, had forgotten the limits to specs (other than flops) wrt exports to China (which may impact some 'stock' product decisions beyond 4090D, perhaps even how certain chips are designed with that limitation in mind in the future) in ways I haven't personally looked into on a deep-enough level to comment. Thanks for the info/reminder.

That said, GDDR would be volatile memory, correct? I imagine (but am not well-versed on the current definition of I/O bw) this is in relation to external bandwidth (links/protocol) and if anything may perhaps only relate to their (capabilities of) external cache structure? While I don't know the actual bw (or limitations of the link), using it's observational impact on performance I've generally equated AMD's L3 structure to something like adding 3mhz(/gbps) over the bus in relation to core speed (as the cache is likely clocked in relation to core; haven't looked into it). So, for instance, if something like a 7800xt/7900gre were running at 2800mhz core clock, the L3 would be contributing what equates to 268.8Gbps for DRAM bw. Obviously it's faster as the size is smaller, but in terms of perf does appear to shake out if you look at it wrt bw limitations. The same is true of nVIDIA's L2 cache and roughly double at same size (similar to core clock*6mhz over the bus), which nVIDIA all-but-confirmed with one product release not so long ago (something in the 4060/4070 series iirc) to excuse the more narrow bus wrt the competition.

In the case of 4090 (with 72MB L2), for instance, that translates to something like ~2730*4.5*384/8 = ~589.68Gbps, which is indeed oddly close to the number you stated. That's internal though, so I don't know if that was part of that decision-making process of the chip and/or effected by those export rules (or rather the rules were built around them after it was completed and/or released)? Now you've got me curious, but I always had a feeling these 'rules' were constructed as a compromise between the government and nVIDIA as for them to still be able use that chip in China, but for them to be conscious of using restraint moving forward.

I'm sure someone more studied could read those last couple paragraphs and declare me a simpleton, and that's fair-enough! I respect there are people that know much more in the intricate areas of these things (and their definitions/classifications) than I do, have have kept up on it better, and/or have information most don't have. I generally focus on the real-world impact to most users, and where limitations are/how they can be improved. I don't always know the correct specific language/terminology regarding how each area is classified. I'm sure there are people around here that can shed much more light on it than myself. I don't *think* what you are implying would apply, but I absolutely could not state that as a fact as I just don't have the inherent knowledge or haven't done the reseach to declare that as fact.
It's possible Hack-A-Day misinterpreted the regulation but what we do know is that the regulation mostly targeted AI/high performance computing which often ship with HBM. The regulation effectively made exporting HBM products to China illegal. The language might be fuzzy but the real world application is the 600 GB/s limit on GDDR/HBM performance which have almost entirely vanished from the Chinese market.
 
Joined
Mar 1, 2021
Messages
414 (0.35/day)
Location
Germany
System Name Homebase
Processor Ryzen 5 5600
Motherboard Gigabyte Aorus X570S UD
Cooling Scythe Mugen 5 RGB
Memory 2*16 Kingston Fury DDR4-3600 double ranked
Video Card(s) AMD Radeon RX 6800 16 GB
Storage 1*512 WD Red SN700, 1*2TB Curcial P5, 1*2TB Sandisk Plus (TLC), 1*14TB Toshiba MG
Display(s) Philips E-line 275E1S
Case Fractal Design Torrent Compact
Power Supply Corsair RM850 2019
Mouse Sharkoon Sharkforce Pro
Keyboard Fujitsu KB955
Keep banging that drum, buddy. I'd love for you and any other people that have similar issues to have them completely resolved. Compatability with old titles can sometimes be a tricky thing (depending on how something was programmed), and sometimes not (it's just a matter of getting ahold of the right person and them having the time to fix it/implement a change). All I can hope is that if there is an easily-enough implementable fix on AMD's side that you are able to make that connection to them and it can/will be sorted to the best of their capability and your appeasement, even if it takes them a few tries to address your (or others) particular issue(s). I wish you much luck, and appreciate you for trying to get that sorted for your community of players. That's what it's all about, right? Just trying to make things the best they can be if it's possible; calling out issues/perceptions and hoping it reaches the right ears so that both they know about it and getting someone capable to help with a solution that is applicable/agreeable to the most people, if able/possible...and doing one's best at going about those things without being offensive/negative as possible.

(The last part is a personal reminder [I mean well, but know I can sometimes come across unintendly harsh towards people when trying to make a point], your attitude towards the issue appears good-natured and on-point, and once-again I applaud your persistance. :))

Yeah I just want it to works like it over a year ago did. Sometimes I tend to get a bit overly engaged and might be misunderstood, but since GSC Gameworld does not care about the game anymore the only company that can help is AMD and I try my very best to get it solved.
 

Feelas

New Member
Joined
Mar 24, 2024
Messages
2 (0.03/day)
The GDDR modules installed on the card are up to the AIB partner's discretion so long as they meet or exceed reference spec. In this case, PowerColor (see OP) has much more capable Samsung chips installed compared your ASRock model. In short, ASRock skimped on the GDDR and PowerColor did not (likely used the same modules the 7900 XT and XTX have). All AMD did was enable overclocking of the VRAM so you can unlock whatever capability the GDDR modules have from the factory that were previously limited by the VBIOS.
I would agree, but we are talking about clocking at 2500MHz and here's the review for Steel Legend.

Qutoing:
The GDDR6 memory chips are made by Samsung and carry the model number K4ZAF325BC-SC20. They are specified to run at 2500 MHz (20 Gbps effective).

So you are suggesting that I have received a different memory PN or Samsung is selling overspecced ICs? Given that TPU BIOS dump for the card is correct, the BIOS supports either Hynix H56G42AS8DX014 or Samsung K4ZAF325BC. Judging by Google, both chip models are possible to find on GRE/XT/XTX all alike and the Samsung part doesn't come in a variant specced for 18Gbp: it is available as SC16 for 16Gbps or SC20 speecced for 20Gbps. Looks like an issue with the memory controller instead, which of course points that it is "in the reference spec", given they underclock for 2250MHz. It is completely baffling they would put 20Gbps VRAMs into a card capable of only 18Gbps...

Perhaps there is a bigger issue for AMD, given that many people on XT/XTX are suffering from black screen issues - did AMD badly bin RDNA3 with modules incapable of >2400MHz VRAM clocks into XT/XTX and those are suffering from the problems? Overclocking on GRE >2400MHz very similarly causes black screen & system hard reset. Maybe there is a bigger RDNA3 failure at play, not a GRE-related one...

LabRat 891 asked for GPU-Z screenshot, here it is.
 

Attachments

  • 7900gre.gif
    7900gre.gif
    31.9 KB · Views: 29
Last edited:

FordGT90Concept

"I go fast!1!11!1!"
Joined
Oct 13, 2008
Messages
26,259 (4.60/day)
Location
IA, USA
System Name BY-2021
Processor AMD Ryzen 7 5800X (65w eco profile)
Motherboard MSI B550 Gaming Plus
Cooling Scythe Mugen (rev 5)
Memory 2 x Kingston HyperX DDR4-3200 32 GiB
Video Card(s) AMD Radeon RX 7900 XT
Storage Samsung 980 Pro, Seagate Exos X20 TB 7200 RPM
Display(s) Nixeus NX-EDG274K (3840x2160@144 DP) + Samsung SyncMaster 906BW (1440x900@60 HDMI-DVI)
Case Coolermaster HAF 932 w/ USB 3.0 5.25" bay + USB 3.2 (A+C) 3.5" bay
Audio Device(s) Realtek ALC1150, Micca OriGen+
Power Supply Enermax Platimax 850w
Mouse Nixeus REVEL-X
Keyboard Tesoro Excalibur
Software Windows 10 Home 64-bit
Benchmark Scores Faster than the tortoise; slower than the hare.
So you are suggesting that I have received a different memory PN or Samsung is selling overspecced ICs? Given that TPU BIOS dump for the card is correct, the BIOS supports either Hynix H56G42AS8DX014 or Samsung K4ZAF325BC. Judging by Google, both chip models are possible to find on GRE/XT/XTX all alike and the Samsung part doesn't come in a variant specced for 18Gbp: it is available as SC16 for 16Gbps or SC20 speecced for 20Gbps. Looks like an issue with the memory controller instead, which of course points that it is "in the reference spec", given they underclock for 2250MHz. It is completely baffling they would put 20Gbps VRAMs into a card capable of only 18Gbps...
It's entirely possible that your specific chip was binned for GRE because the memory controller couldn't handle 7900 XT spec.



I did a lot more digging into what happened. Short version is this:
1) November 2022, the performance export ban was issued that mentions the 600 GB/s rule on non-volatile memory: https://hackaday.com/2022/11/09/chi...cially-slowed-to-dodge-us-export-regulations/
2) July 2023, 7900 GRE launches: https://www.tomshardware.com/news/amd-radeon-rx-7900-gre-launch
3) October 2023, Department of Commerce announces change in the export ban that strongly targets tensor cores (TFLOP calculation): https://www.tomshardware.com/news/no-nvidia-isnt-breaking-gpu-sanctions-analyst
4) November 2023, the rule goes into effect, 4090 disappears because it's not compliant, and 7900 series sales surge in China to fill the demand: https://hothardware.com/news/amds-flagship-radeon-rx-7900-xtx-and-xt-gpus-flourish-in-china

I believe GRE was created in case the rule was modified to apply the 600 GB/s rule to volatile memory which would limit Chinese sales of 7900 XT and 7900 XTX. Dell apparently misinterpreted the rule and proactively applied it in November 2023: https://www.techpowerup.com/316044/...high-end-radeon-and-instinct-mi-gpus-in-china

AMD wasn't the only company to release a 600 GB/s product preemptively; Birin Technology (a Chinese company) did the same (see hackaday article above).

Because the 2023 replaced the rule invalidating the 2022 rule which mentioned 600 GB/s, AMD felt they were no longer threatened by it so they enabled overclocking in the driver for the card.

So, all good now, except that Chinese can't get 4090 anymore (hence 4090D https://www.tomshardware.com/pc-com...nt-model-has-fewer-cores-and-lower-power-draw).
 
Joined
May 13, 2008
Messages
723 (0.12/day)
System Name HTPC whhaaaat?
Processor 2600k @ 4500mhz
Motherboard Asus Maximus IV gene-z gen3
Cooling Noctua NH-C14
Memory Gskill Ripjaw 2x4gb
Video Card(s) EVGA 1080 FTW @ 2037/11016
Storage 2x512GB MX100/1x Agility 3 128gb ssds, Seagate 3TB HDD
Display(s) Vizio P 65'' 4k tv
Case Lian Li pc-c50b
Audio Device(s) Denon 3311
Power Supply Corsair 620HX
I did a lot more digging into what happened. Short version is this:
1) November 2022, the performance export ban was issued that mentions the 600 GB/s rule on non-volatile memory: https://hackaday.com/2022/11/09/chi...cially-slowed-to-dodge-us-export-regulations/
2) July 2023, 7900 GRE launches: https://www.tomshardware.com/news/amd-radeon-rx-7900-gre-launch
3) October 2023, Department of Commerce announces change in the export ban that strongly targets tensor cores (TFLOP calculation): https://www.tomshardware.com/news/no-nvidia-isnt-breaking-gpu-sanctions-analyst
4) November 2023, the rule goes into effect, 4090 disappears because it's not compliant, and 7900 series sales surge in China to fill the demand: https://hothardware.com/news/amds-flagship-radeon-rx-7900-xtx-and-xt-gpus-flourish-in-china

I believe GRE was created in case the rule was modified to apply the 600 GB/s rule to volatile memory which would limit Chinese sales of 7900 XT and 7900 XTX. Dell apparently misinterpreted the rule and proactively applied it in November 2023: https://www.techpowerup.com/316044/...high-end-radeon-and-instinct-mi-gpus-in-china

AMD wasn't the only company to release a 600 GB/s product preemptively; Birin Technology (a Chinese company) did the same (see hackaday article above).

Because the 2023 replaced the rule invalidating the 2022 rule which mentioned 600 GB/s, AMD felt they were no longer threatened by it so they enabled overclocking in the driver for the card.

So, all good now, except that Chinese can't get 4090 anymore (hence 4090D https://www.tomshardware.com/pc-com...nt-model-has-fewer-cores-and-lower-power-draw).

Appreciate your time on the research; I figured it was likely links (chip-to-chip communication; interconnects: ex infinity fabric et al) that could be used for things other than just volatile memory.

Yeah, 4090 was a no-no due to FP32 (Tflops), I knew that part. Didn't know/remember there was a 8-bit (TOP) performance(/density) update to the law, but I can certainly believe that part (I won't get into it).

I think we used to live in a world that largely relied on full/double precision (FP32/64), where-as now that has morphed into larger use of FP16 and 8-bit calculations (for which can be more densely packed).

I really should read the actual current guidelines, as obviously and as you can see, things can be easily misinterpreted and incorrect information passed along (even by the press and/or major companies).

It's actually quite fascinating imho: It used to be these companies were so far ahead of the governments they would circumvent whatever (eventually) archaic rule by implementing different design decisions. In this case the government actually appeared to very much understand what they were doing (at least in the updated law, which makes sense) while perhaps some in the supply chain did not (or keep current).

That said, I quite dislike talking about this stuff, tbh. Although it is important to understand the what/why, I personally very much prefer to be a unifier rather than discussing limitations in international trade.

Edit: TMW you look at your post later and realize you meant double-precision (64-bit floating point), single precision (FP32), and half-precision (FP16)...but worded it incorrectly.
 
Last edited:
Top