Hi,
I don't know where to turn with this as EVGA support seem to run around in circles and not read cases properly.
Here is the dialogue so far with their support arm (greetings removed).... what do I do next?... I'm back at square one with a duff card!
16/11/2014 Logged call with Nvidia via live chat for Black screen at boot after windows logo on Display Port to Samsung 4k (590D) monitor.
Nvidia support hooked me up via telephone to EVGA support who logged this under EVGA Ticket: 1816882 with an action to resolve with a new BIOS.
They took my serial number at this stage. 18/11/2014 Nvidia support sent me a ‘hot fix driver which resolved the issue temporarily,
but after a further driver update, the issue came back. 24/11/2014 EVGA support asked me to rate their support via email.
06/01/2015 EVGA provided a fixed BIOS via a vague email with no instructions: Here is a link to a new BIOS to test, this should resolve the issues with your display port.
06/01/2015 I emailed EVGA for more info: nVidia sent me a hot fix for 344.80 drivers… but it broke again with the latest whql drivers.
What do I need to do to get this updated?
And what guarantees have you got that it will be OK?
… how do I revert back if the ‘test’ does not workout?
06/01/2015 EVGA responded with more information:
EMEA Support Agent #1- If you have not done so re-run the BIOS update however the drivers would not adjust the card at that level. If you have not done so try reloading the drivers with the clean install option enabled under Custom after the drivers uncompress. If there is no change you can uninstall the drivers then reload the earlier 344.80 drivers and try the next drivers once they are released. There should not be any issues with changing drivers as they are not writing to the card and as long as there is not a power failure our windows issues flashing the card again would not be an issue. Due to possible windows or power grid issues happening at the same time as the flash we do recommend to minimize the number of flashes and only do it when the windows and power are completely stable, IE not crashing frequently or during a storm.
19/01/2015 I updated the BIOS (I had been overseas for 10 days), but noticed there was now a lot of crashes, so contacted EVGA support: The BIOS fixed the black screen after the windows logo on windows 8.1 with the display port/Samsung 4k monitor, however I now have frequent crashes in-game and the TDP seems to be going higher than it was before. Can you advise on what to change in line with the BIOS fixes you have provided.. I have updated to the latest 347.09 driver from nVidia.
20/01/2015 EVGA responded:
EMEA Support Agent #2– thanks for your detailed request - did you handle a clean driver installation by using any kind of open source tools like driver fusion first, before re-installing the latest - would be recommended regarding the registry entries - please let us know first.
20/01/2015 I responded via email: I used DriverCleaner.
22/01/2015 EVGA responded:
EMEA Support Agent #3- there is a thread about an issue like this in the geforce forums, for different users different manually settings for the color and disaply input did help to fix the problem. Have you already seen this and tried those suggestions, this can probably help to get it to work.
22/01/2015 Losing patience, I responded with a lot of detail:
I’m already on that thread from when I had a Samsung 4K TV (55” curved) but that thread is about Chroma (4:4:4 vs 4:2:0) and the processor in the TV is actually not enabled in the TV Firmware of some of the product lines. Samsung has actually updated the firmware for most their TVs now so all should be OK… I was in direct contact with Nvidia and Samsung at that point.Anyhow, the thread on the GeForce forums is irrelevant to the problem I am experiencing entirely and I have worked out what sequence of events is causing the crashes.
The crashes are happening when the GPU voltage drops below 1.212 along with a core frequency drop whilst in a 3D game (not stereo 3D, just a normal non-desktop full screen game). This usually happens when the GPU and it’s surrounding chips is not being worked hard enough (e.g. low AA or IQ settings)… similar to P-States on CPUs I assume. This voltage/freq drop in turn is not catered for in the game’s graphics engine and so causes the abend (grey screen or driver crash depending on game) either way the PC needs restarting to resolve the issue, either a hard reset for the grey screen or a clean shut down for the driver crash.
This did not happen noticeably before the BIOS update on driver 344.80 or 347.09 (when I could get the latter to boot in the first place).
I have reduced the amount of crashes by running k-boost (part of the EVGA Precision X software that applies max performance GPU Boost constantly), but the voltage/frequency of the GPU clock still goes down marginally when the GPU is not being taxed enough despite the GPU not dropping below 99% utilisation throughout (again I assume this is some form of p-state behaviour). I have the SC version of the 980 so it is overclocked to start with when it leaves your factory. If I reduce the core freq by an offset of -300, the crashes never occur and the frequency and voltage remains perfectly stable, but this is not why I bought a SC card and paid the premium for it.
There are many forum entries discussing this behaviour of the 980 and the 970 from EVGA and MSI and it seems it can be overcome with a custom BIOS tweak that forces the GPU boost feature off and the frequency up and also some hardware modifications that address the same issues, but I really don’t want to go down either of those routes to invalidate the warranty.
Hopefully with this information you can start to investigate how to correct your BIOS and/or work with Nvidia to change the behaviour in the Driver/reference BIOS so that the GPU Voltage remains constant when in 3D mode and the core frequency doesn’t have the drops when not taxed enough while maintaining the available power (TDP) at just under the maximum permitted. It also looks to be a characteristic of early cards like mine and not common on the later versions so something may have been changed that you can release to me or swap my card for one that doesn’t have the problem.
23/01/2015 EVGA email:
EMEA Support Agent #3- thanks for all the information and explanation about this. I am not sure if it is possible to fix the voltage on a stable level because the chip is always acting individually depending on the load. That what the K-boost feature is ment to be, this should normally fix it to a stable level for issues like this in game. I will forward this information to our product management Team but not sure if this will help because it looks like other screens do not have this issue. I would still also like to do an RMA on this card, maybe there is a problem with the voltage regulation on your card and another card will work fine for this with the Kboost feature. We will do the RMA cost free for you, can you confirm for this an address where UPS can pick up the card? Because everything is working fine with less clock speed it do sound to me like something is not running right within the card.
23/01/2015 My email response: The problem with the screen being black at boot is resolved with the latest BIOS you have provided.
The symptoms I have described below are true on any screen, you are confused. I have 3 screens, the Samsung 4k, a 1080p 8” screen and a BenQ 1080p 3D projector…. I can choose any of these as the primary and only display and still get the voltage/frequency drops in situations where the GPU is not being taxed enough along with countless others that are experiencing the same issue on many forums.
The RMA is a great idea, but only if later cards that you provide have this issue addressed in the Hardware or Firmware and I would need an advanced replacement as this is also my work PC and I could not do without it.
Please let me know if you can help me out considering the situation I describe.
Please ignore this in the code box as the test was invalid
25/01/2015 I performed some more tests to see if I could see what was making it crash so aften and emailed the info to EVGA: Further to my last email, I seem to be suffering from yet another widespread problem with my GTX980 SC.
Memory allocation above 3.5GB is bandwidth limited…. Could this be a reason why I’m crashing so frequently?
The problem seems to affect all 970’s and some early 980’s… and mine was one of the first out of the door.
See attached Results.txt. You will notice that the final 3 allocations on each of the tests, the bandwidth is pulled right back.
This is just not acceptable for the price paid for this card.
Results.txt:
Nai's Benchmark
Allocating Memory . . .
Chunk Size: 128 MiByte
Allocated 30 Chunks
Allocated 3840 MiByte
Benchmarking DRAM
DRAM-Bandwidth of Chunk no. 0 (0 MiByte to 128 MiByte):176.65 GByte/s
DRAM-Bandwidth of Chunk no. 1 (128 MiByte to 256 MiByte):176.41 GByte/s
DRAM-Bandwidth of Chunk no. 2 (256 MiByte to 384 MiByte):177.00 GByte/s
DRAM-Bandwidth of Chunk no. 3 (384 MiByte to 512 MiByte):176.37 GByte/s
DRAM-Bandwidth of Chunk no. 4 (512 MiByte to 640 MiByte):176.71 GByte/s
DRAM-Bandwidth of Chunk no. 5 (640 MiByte to 768 MiByte):176.71 GByte/s
DRAM-Bandwidth of Chunk no. 6 (768 MiByte to 896 MiByte):176.53 GByte/s
DRAM-Bandwidth of Chunk no. 7 (896 MiByte to 1024 MiByte):176.85 GByte/s
DRAM-Bandwidth of Chunk no. 8 (1024 MiByte to 1152 MiByte):176.08 GByte/s
DRAM-Bandwidth of Chunk no. 9 (1152 MiByte to 1280 MiByte):176.49 GByte/s
DRAM-Bandwidth of Chunk no. 10 (1280 MiByte to 1408 MiByte):177.04 GByte/s
DRAM-Bandwidth of Chunk no. 11 (1408 MiByte to 1536 MiByte):176.52 GByte/s
DRAM-Bandwidth of Chunk no. 12 (1536 MiByte to 1664 MiByte):176.76 GByte/s
DRAM-Bandwidth of Chunk no. 13 (1664 MiByte to 1792 MiByte):176.91 GByte/s
DRAM-Bandwidth of Chunk no. 14 (1792 MiByte to 1920 MiByte):176.55 GByte/s
DRAM-Bandwidth of Chunk no. 15 (1920 MiByte to 2048 MiByte):176.89 GByte/s
DRAM-Bandwidth of Chunk no. 16 (2048 MiByte to 2176 MiByte):176.18 GByte/s
DRAM-Bandwidth of Chunk no. 17 (2176 MiByte to 2304 MiByte):176.57 GByte/s
DRAM-Bandwidth of Chunk no. 18 (2304 MiByte to 2432 MiByte):176.96 GByte/s
DRAM-Bandwidth of Chunk no. 19 (2432 MiByte to 2560 MiByte):176.58 GByte/s
DRAM-Bandwidth of Chunk no. 20 (2560 MiByte to 2688 MiByte):176.69 GByte/s
DRAM-Bandwidth of Chunk no. 21 (2688 MiByte to 2816 MiByte):176.96 GByte/s
DRAM-Bandwidth of Chunk no. 22 (2816 MiByte to 2944 MiByte):176.54 GByte/s
DRAM-Bandwidth of Chunk no. 23 (2944 MiByte to 3072 MiByte):176.84 GByte/s
DRAM-Bandwidth of Chunk no. 24 (3072 MiByte to 3200 MiByte):176.72 GByte/s
DRAM-Bandwidth of Chunk no. 25 (3200 MiByte to 3328 MiByte):176.52 GByte/s
DRAM-Bandwidth of Chunk no. 26 (3328 MiByte to 3456 MiByte):176.96 GByte/s
DRAM-Bandwidth of Chunk no. 27 (3456 MiByte to 3584 MiByte):28.36 GByte/s
DRAM-Bandwidth of Chunk no. 28 (3584 MiByte to 3712 MiByte):15.34 GByte/s
DRAM-Bandwidth of Chunk no. 29 (3712 MiByte to 3840 MiByte):12.23 GByte/s
Benchmarking L2-Cache
L2-Cache-Bandwidth of Chunk no. 0 (0 MiByte to 128 MiByte):515.34 GByte/s
L2-Cache-Bandwidth of Chunk no. 1 (128 MiByte to 256 MiByte):515.41 GByte/s
L2-Cache-Bandwidth of Chunk no. 2 (256 MiByte to 384 MiByte):515.31 GByte/s
L2-Cache-Bandwidth of Chunk no. 3 (384 MiByte to 512 MiByte):515.37 GByte/s
L2-Cache-Bandwidth of Chunk no. 4 (512 MiByte to 640 MiByte):515.38 GByte/s
L2-Cache-Bandwidth of Chunk no. 5 (640 MiByte to 768 MiByte):515.48 GByte/s
L2-Cache-Bandwidth of Chunk no. 6 (768 MiByte to 896 MiByte):515.49 GByte/s
L2-Cache-Bandwidth of Chunk no. 7 (896 MiByte to 1024 MiByte):515.38 GByte/s
L2-Cache-Bandwidth of Chunk no. 8 (1024 MiByte to 1152 MiByte):515.34 GByte/s
L2-Cache-Bandwidth of Chunk no. 9 (1152 MiByte to 1280 MiByte):515.33 GByte/s
L2-Cache-Bandwidth of Chunk no. 10 (1280 MiByte to 1408 MiByte):515.41 GByte/s
L2-Cache-Bandwidth of Chunk no. 11 (1408 MiByte to 1536 MiByte):515.44 GByte/s
L2-Cache-Bandwidth of Chunk no. 12 (1536 MiByte to 1664 MiByte):515.50 GByte/s
L2-Cache-Bandwidth of Chunk no. 13 (1664 MiByte to 1792 MiByte):515.54 GByte/s
L2-Cache-Bandwidth of Chunk no. 14 (1792 MiByte to 1920 MiByte):515.44 GByte/s
L2-Cache-Bandwidth of Chunk no. 15 (1920 MiByte to 2048 MiByte):515.31 GByte/s
L2-Cache-Bandwidth of Chunk no. 16 (2048 MiByte to 2176 MiByte):515.46 GByte/s
L2-Cache-Bandwidth of Chunk no. 17 (2176 MiByte to 2304 MiByte):515.33 GByte/s
L2-Cache-Bandwidth of Chunk no. 18 (2304 MiByte to 2432 MiByte):515.58 GByte/s
L2-Cache-Bandwidth of Chunk no. 19 (2432 MiByte to 2560 MiByte):515.31 GByte/s
L2-Cache-Bandwidth of Chunk no. 20 (2560 MiByte to 2688 MiByte):515.42 GByte/s
L2-Cache-Bandwidth of Chunk no. 21 (2688 MiByte to 2816 MiByte):515.53 GByte/s
L2-Cache-Bandwidth of Chunk no. 22 (2816 MiByte to 2944 MiByte):515.38 GByte/s
L2-Cache-Bandwidth of Chunk no. 23 (2944 MiByte to 3072 MiByte):515.45 GByte/s
L2-Cache-Bandwidth of Chunk no. 24 (3072 MiByte to 3200 MiByte):515.49 GByte/s
L2-Cache-Bandwidth of Chunk no. 25 (3200 MiByte to 3328 MiByte):515.41 GByte/s
L2-Cache-Bandwidth of Chunk no. 26 (3328 MiByte to 3456 MiByte):515.38 GByte/s
L2-Cache-Bandwidth of Chunk no. 27 (3456 MiByte to 3584 MiByte):14.78 GByte/s
L2-Cache-Bandwidth of Chunk no. 28 (3584 MiByte to 3712 MiByte):14.78 GByte/s
L2-Cache-Bandwidth of Chunk no. 29 (3712 MiByte to 3840 MiByte):18.80 GByte/s
26/01/2015 EVGA Responded:
EMEA Support Agent #3- by the attachement it do look like there is an issue like this. So far we did not had any customer that did face that issue with GTX 980s, only at GTX 970s.
Is there a NVIDIA or EVGA branding right on top of the PCI.E Slot?
We do offer an advanced RMA service but for this you need to place a security collateral via paypal. If you are willed to do this a replacement in advance is no problem from our site.
Can you let me know the S/N of the problem to take a look into this?
26/01/2015, my response: Not sure what you mean by: “Is there a NVIDIA or EVGA branding right on top of the PCI.E Slot?”
Could you elaborate? My motherboard (which has the PCI.E slots that the EVGA Card fits into) is an ASUS X99-S, so it won’t have any branding from you… unless you are secretly the same company…. So what exactly do you mean?
The serial number is registered with you my friend, you already have this information.
What do you mean by place a security collateral by PayPal? How much… ? Does this remain a ‘pending transaction’ or does it take money from my account straight away regardless?
Please be clear with your responses to avoid any further delay or misunderstandings.
27/01/2015 EVGA:
EMEA Support Agent #3- sorry, I mean the branding at the graphics card. At the front site of the graphics card is above the PCI.E connector a branding, this is a EVGA or NVIDIA logo. This info is to get to know who produced the card.
The collateral for this is the new price of the card, the reason for this is if we do not get the old card back we just handle it like anormal purchase. If you get charged right away with this is depending on how you have setup your paypal account, if this is connected to a bank account it will charge the account right away. If you have it connected with a credit card the account will charge the card just at the end of the monthly periode. If we handle the RMA and refund you before the end of the periode the credit card will not get charged.
27/01/2015 Me:OK, thanks for the explanation.
I can’t afford to pay for another graphics card, they are not cheap as this is almost the top of the range card, can you please arrange collection and a full refund. I will purchase another brand once you give me the money.
28/01/2015 EVGA:
EMEA Support Agent #4- thank you for your email. Please note, if we have not received any account information from you we are not able to reference the account to effect any changes. For any future requests please reference your serial number for the fastest response.
For the interm, our Terms and Conditions regarding Warranty/Refunds/Exchanges can be found on our website here:
28/01/2015 Me: I bought the goods (SN 1410732982601249) in the UK, therefore I am covered by the sale of goods act and the trade description act both of which can be applied to this product which does not do what it is supposed to (Memory is defective after 3.5GB and it crashes due to a throttling problem too), therefore your terms and conditions are overridden entirely and I am asking for a refund in full as the product is not as described.
Please supply details of your legal representative.
29/01/2015 EVGA: EMEA Support Agent #4- thank you for your email and included details. For products bought through resellers, the customer needs to return the product to the same reseller. To address the memory issue that you mention, I have included the following details below:
Thank you for being an EVGA customer for your inquiry on the GTX 970 graphics card. We are very sorry to hear of the recent confusion on how the memory on the GTX 970 works.
I assure you however that here at EVGA we stand behind our products and if you are unhappy with your product then we can offer you a one time special option to Step-up to a GTX 980 card by using our optional Step-up program. Please visit for more details on how this program works and the regions that it is available in.
In order to try and assist further we have included some performance data below so that you can see actual testing results between the GTX 970 and the GTX 980.
Shadows of Mordor: GTX 980 GTX970
<3.5GB settings: 2688x1512 Very High 72FPS 60FPS
>3.5GB settings: 3456x1944 Very High 55FPS (24%) 45FPS (-25%)
Relative performance difference: 1%
Battlefield 4:<3.5GB settings: 3840x2160 2xMSAA 36FS 30FPS
>3.5GB settings: 3840x2160 135% res. 19FPS (-47%) 15FPS (-50%)
Relative performance difference: 3%
Call of Duty: Advanced Warfare:<3.5GB settings: 3840x2160 FSMAA T2x, SS off 82FPS 71FPS
>3.5GB settings: 3840x2160 FSMAA T2x, SS on 48FPS (-41%) 40FPS (-44%)
As you can see above the relative performance difference varies from 1% in Shadows of Mordor, to 3% in Battlefield 4 or Call of Duty: Advanced Warfare. The performance differences between a 970 and 980 in the above examples remain a constant 15-19%, regardless of memory usage.
I hope that the above information helps clarify the performance from the GTX 970. Please let us know if you do want to continue with a step-up on your product for the higher performing GTX 980 card, and we'll get that option opened up for you.
29/01/2015 Me: FOR GODS SAKE MAN!!!
I HAVE A GTX 980 SC already… that’s the card with the issue…
Heeeeeeelllllp! (please)
post edited by MrPix - 2015/02/06 01:31:10