GeForce GTX 970 Memory Issue Fully Explained - Nvidia's Response

There have been numerous reports of the GTX 970 suffering from a retention allocation issue limiting information technology to 3.5GB instead of the full 4GB. We've done our ain testing and analysis of this issue here at Wccftech.com in a previous slice. And in this article we aim to explain the issue in totality, why it occurs, what information technology means and how it affects users.

NVIDIA GM204 GPU Powering the GTX 980 and GTX 970 This whole debacle started when GTX 970 users began reporting an issue with retentiveness allocation on their cards. The event specifically was effectually the GTX 970 not being able to address all of the 4GB of on-board GDDR5 memory that Nvidia has equipped its cards with. GTX 970 users continuously reported maximum memory usage hitting a wall at 3.5GB. Shortly afterwards nosotros went alee and tested out the GTX 970 to endeavour and meet where this upshot is stemming from. What we had discovered was quite interesting and goes hand in mitt with Nvidia's latest response to this issue. But earlier we discuss our findings let's see what Nvidia has to say first.

Nvidia Responds to GTX 970 Retentiveness Allotment Outcome

In a recent argument to PCPer.com Nvidia admitted the beingness of a retentiveness allocation behavior that led so many GTX 970 users to worry. The company as well offered a brief caption of the phenomenon and why it occurs.

"The GeForce GTX 970 is equipped with 4GB of dedicated graphics retentiveness. Yet the 970 has a different configuration of SMs than the 980, and fewer crossbar resources to the memory system. To optimally manage retention traffic in this configuration, nosotros segment graphics memory into a three.5GB section and a 0.5GB section. The GPU has higher priority access to the iii.5GB section. When a game needs less than three.5GB of video memory per draw command then information technology will only access the first partitioning, and 3rd party applications that measure memory usage will written report iii.5GB of retentivity in employ on GTX 970, but may report more for GTX 980 if there is more memory used by other commands. When a game requires more iii.5GB of retentiveness then we employ both segments.

Nosotros understand in that location accept been some questions about how the GTX 970 will perform when it accesses the 0.5GB memory segment. The best style to test that is to look at game performance. Compare a GTX 980 to a 970 on a game that uses less than three.5GB. Then plough up the settings so the game needs more than 3.5GB and compare 980 and 970 performance over again.

Hither's an example of some performance data:

	GTX 980	GTX 970
Shadow of Mordor
<3.5GB setting = 2688x1512 Very Loftier	72 FPS	60 FPS
>3.5GB setting = 3456x1944	55 FPS (-24%)	45 FPS (-25%)
Battlefield 4
<3.5GB setting = 3840x2160 2xMSAA	36 FPS	xxx FPS
>iii.5GB setting = 3840x2160 135% res	nineteen FPS (-47%)	15 FPS (-l%)
Telephone call of Duty: Advanced Warfare
<3.5GB setting = 3840x2160 FSMAA T2x, Supersampling off	82 FPS	71 FPS
>iii.5GB setting = 3840x2160 FSMAA T2x, Supersampling on	48 FPS (-41%)	40 FPS (-44%)

On GTX 980, Shadows of Mordor drops about 24% on GTX 980 and 25% on GTX 970, a 1% divergence. On Battlefield iv, the drop is 47% on GTX 980 and 50% on GTX 970, a three% departure. On CoD: AW, the drop is 41% on GTX 980 and 44% on GTX 970, a 3% difference. Equally you can run across, in that location is very lilliputian change in the performance of the GTX 970 relative to GTX 980 on these games when it is using the 0.5GB segment."

Nvidia'southward argument explains our ain findings earlier. In our analysis we discovered that the GTX 970 tin can actually address and employ all available 4GB of VRAM and non merely 3.5GB. Nevertheless the performance penalization for using that final 0.5GB of VRAM is still interesting. As Nvidia explains it and as we had suspected earlier, the issue is essentially due to the fact that the GTX 970 is based on a cutting-down GM204-200 GPU die instead of the full GM204 flake similar its bigger blood brother the GTX 980.

The GM204-200 GPU has three SMM units disabled, for a total of 13/16 SMMs. This in plough disables some components of the memory sub-system. This leaves the chip with less resource to manage the aforementioned amount of retentivity as the GTX 980. Which is why Nvidia resorted to the dual segment (three.5GB-0.5GB) retentivity arrangement. This in plough insures that the larger 3.5GB segment doesn't endure any operation penalty.

Nvidia doesn't provide data to compare how a 970 would perform had it non exhibited this memory behavior. So I went alee and calculated how the operation of the GTX 970 would've been affected had this retention allocation phenomenon was absent. I've done this by applying the performance scaling ratio of the GTX 980 ,which doesn't exhibit this behavior, to the GTX 970.

Wccftech	GTX 970 Operation in %	GTX 970 Functioning Without Retentivity Allocation Issue in %
Shadow of Mordor	100%	102%
Battleground 4	100%	105%
Telephone call of Duty Advanced Warfare	100%	104%

Simply put the GTX 970 is simply able to fully utilize 3.5GB out of the 4GB of bachelor VRAM optimally. While the carte du jour can nonetheless access all 4GB of VRAM, the last 0.5GB of memory is not accessed or managed as efficiently as the residue of the available memory. Which leads to the reported performance degradation when an application needs to access more than than 3.5GB. In the examples Nvidia provided to the press the performance penalty is in the mid single digit percentages. Notwithstanding we don't fully know withal if in other usage scenarios the retentiveness issue volition show to be more than detrimental. Equally there's bear witness that it has a more than perceptual impact on frame fourth dimension consistency rather than FPS which can be perceived as the stuttering or choppiness reported by Nvidia users on the GeForce forums.

Well-nigh users should not face this outcome except on higher resolutions such as 4K or in the instance of multi-monitor gaming. In which case we would actually recommend buying a graphics card with more than memory anyway. AMD already has 8GB R9 290X cards in the market which were introduced iii months agone. Earlier rumors indicated that Nvidia was besides launching 8GB variants of the GM204 based 900 serial cards in November of terminal year but that unfortunately did non happen. Samsung SDRAM schedules indicate that Nvidia could launch 8GB variants of their 900 serial cards this quarter.

Although with rumors looming of a GM200 launch taking place in ii months time and AMD'due south HBM equipped Fiji XT coming shortly after. I'd doubtable that users would want to meet how well products based on these new GPUs perform in the market and so make a decision.