Area 51 upgrade to Asus R9 290X DCII Problem

oops, here's the original post:

http://en.community.dell.com/owners-club/alienware/f/3746/t/19563242.aspx?pi21932=3#20579083

In this post, we decided the motherboard needs an evga power boost, to give a kick to the pci-e lane:

• http://www.frozencpu.com/products/22593/cab-1928/EVGA_PCI-E_Power_Boost_Cable_100-MB-PB01-BR.html

On page two, the user purchased a pair of "y-split" cables, in tandem w/pwr boost:

search term, eBay = evga adapter:

• http://www.ebay.com/sch/Cables-Connectors-/31491/i.html?_from=R40&_nkw=evga+adapter&_sop=15

example vendors, $8 each, 6pin to dual-6pin cable:

• http://www.ebay.com/itm/EVGA-6-Pin-Power-Adapter-Converter-Connector-Video-Card-Graphics-New-in-Package-/131133293510?pt=US_Power_Cables_Connectors&hash=item1e882743c6
• http://www.ebay.com/itm/EVGA-6-Pin-Power-Adapter-Converter-Connector-Video-Card-Graphics-/271512112155?pt=US_Power_Cables_Connectors&hash=item3f37620c1b

On page three, the use of the optional 1200w unit, VHM5V, was suggested. If your psu is model j297r, 99% chance it will need the pwr boost & a pair of y-splits, which are for single card useage. When running dual cards, like 7990/290x, boxing j297r for a psu swap is suggested; 1200 watts yes, but only rated @ 23.26 amps per card due to over-current-protection  ... Good luck

So I was wrong, I do NOT have a 1200w power supply, I have the 1000W power supply model H1000E-01.  This model only has 2 PCIE cables, with 4 total connectors P16/17/18/19.  The label is as follows:

+5V 32A             +3.3V 30A

+12VA 18A        +12VB 18A

+12VC 18A       +12VD 18A

+12VE 18A

-12V .2A             +5VFP 3A

I am assuming that each PCI-E cable is on 1 of the 12V rails with 18A each, so wouldn't that satisfy the 12v/24A requirement of the card?

I tried to run a few benchmark runs on the R9 290X of Tomb Raider and logged the results with GPU-Z and found that the card was never drawing more than 205Watts.  The reading on the 12V sensor under load is 11.5V, and the GPU is utilized at 100% until crash.  I went ahead and put my HD5970 back in and it is still running fine.

As a note, I was sure that I ordered the 1200W PSU, but I can't find any mention of what PSU I ordered anywhere on the invoice....Anyhow...

Any thoughts on how I might proceed?

And thanks again!

Yes, mthrbrd pci-e lane may be starved, so order the evga pwr boost; if psu still shuts down, either fall back on the 5970, or replace w/j297r (& y-splits if needed) or vhm5v. The 1100w unit is 18a also, you only get one more cable/rail out of it, so nix the 1100w idea, but that's my opinion.

You don't sound ready for a total psu swap yet, but you got short-changed on the 1Kw unit. If you want to treat your rig to a new psu, find me on eBay, search corsair 1200 or seasonic 1250, you'll see me & my stuff there ...

I want to explain a little more clearly what is actually happening, because the power supply is not actually shutting down.  What happens is within 5-10 seconds of putting a 3D (gaming) load on the GPU (Tomb Raider benchmark), the screen goes black, the sound continues for several seconds before also getting "stuck" (usually stuttering on the last bit of sound that was played) and the computer is locked up.  I have a Logitech G19 keyboard, and I leave the clock up on the LCD, and after the screen goes black, the clock second hand keeps moving until the sound stops and the clock stops too.  The system is still on (fans running, AlienFX is on, HDDs are spinning etc.) and at this point the only thing to do is hold the power button for a hard reset.

I have my entire rig, including my NAS, monitor, the aforementioned Logitech G19, and my 5.1 speaker system, connected to a 1200W BackUPS, and it has a wattage monitor.  The peak wattage it shows during the Tomb Raider benchmark is less than 600W peak, even with all of this equipment attached to it.  To eliminate the BackUPS as a problem, I disconnected the PC from the BackUPS and connected it straight into an electrical outlet that is on a totally separate circuit from the rest of my PC equipment (I ran 2 circuits to my office when I wired it) and the results were the same.

I did some logging with GPU-Z during the Tomb Raider benchmark, and it never logs amp usage higher than 20A.  I also tried Arkham Origins benchmark, but it won't even get past the loading screen.  Despite my original post, it doesn't seem to make any difference which PCI-E connectors I am using, the time to lockup on the Tomb Raider benchmark is variable between 5-10 seconds, and the Arkham Origins benchmark won't get past the loading screen.  Other than the GPU, the only items connected to my motherboards PCI connectors are a USB 3.0 card and a X-FI Titanium Sound card.  I tried running the benchmarks with the USB 3.0 card disconnected but it didn't make any difference.

I have no problems running either of these benchmarks (or any other games/benchmarks) on my HD5970, which is currently using only the P16/P17 connectors which are on one PCI-E cable.

I have been reading on various forums, and opinions on the different R9 290/290X "black screen issue" vary from being bad Elpidia GDDR ram on various models of the R9 290X including the Asus R9 290X DCII, to gpu voltage issues, bios issues, and driver issues, and power supply issues.  Some claim that the card is just drawing too much power for lower-amped 12v rails.  The card itself says it requires a power supply capable of providing 12V/24A.  One person has even concluded that the issue is a poor DVI connection that is causing the error, and his description of what he is experiencing is identical to what I am experiencing.

Anyway, I ordered the EVGA Power Booster since it was pretty inexpensive on Amazon.com, so once that arrives I will give it a test to see if it makes a difference; Thanks Cass :)  Perhaps while I wait on that you (or anyone else) might be able to give me some advise on the PSU in my Area-51:

I have the 1000W model H1000E-01 which is also Dell P/N 0U662D

Here is a screen shot of the lable...

From the label, I gather that this PSU has 5x 12V rails @ 18A/rail.  What I am curious about is which 12V rails the 2 PCI-E power cables (with connectors P16/17 and P18/19) might be on.  Mainly I want to know if each PCI-E cable is on its own rail, if the 2 cables are one 1 rail, or if either cable shares a rail with other hardware.  If anyone knows this answer, or can advise me on what I would need to do to test it and find out, I would appreciate that.

Thanks again!

Thanks Cass-Ole, the PCI-E cables are Yellow and Blue/Yellow, and I don't think I will be tearing my PSU apart to investigate it any further :)  I am am pretty sure now that the GPU is defective, and have RMAed it.  I believe, based off of the information on the PSU label and from your information on the PCI-E cables being on their own rails, that the single R9 290X should receive plenty of power from the 1000W PSU (18Ax2=36A total from the PCI-E power cables plus the PCI-E slots themselves).  The power requirements are close between the HD5970 and the R9 290X, and I haven't had any trouble with my HD5970.  If my replacement GPU still has problems, I have the EVGA power boost on the way to test, plus I think I can get my hands on a 750W PSU to try running the card off of a separate supply from the system PSU.

Thanks again for the help, I will post a reply once the replacement card arrives and I have had a chance to do some more tests :)

So my replacement R9 290X came today, and it is rock solid.  The original obviously was defective.  Thanks again for your help Cass-Ole :)

So my replacement R9 290X came today, and it is rock solid.  The original obviously was defective.

What brand was it? Was it new? I'll make a note to avoid them.

What brand was it? Was it new? I'll make a note to avoid them.

ASUS manufactured AMD R9 290X DCII-OC.  I purchased it new, and the company that sold it to me RMAed it and sent a new replacement that arrived today and (so far) is working great.  I have been benchmarking it against my HD5970; 6 gaming benchmarks using maxed settings (Batman: Arkham Origins, Bioshock Infinite, Hitman: Absolution, Metro: Last Light, Thief, Tomb Raider, and Total War: Rome II) and I am getting +30% to +60% FPS @1080P.  Also, there won't be any CrossFireX "microstutter" from the HD5970 since this is just a single core GPU, which is nice.

The ASUS R9 290X is generally rated well, but there are reports not only of the ASUS version of the R9 290X but of various vendors models of the R9 290X and R9 290 having various black screen issues, and various theories as to why they occur.  What was happening in my case was that the card would run fine in normal tasks like browsing the internet, sitting at the desktop etc., but within 10 seconds of putting a gaming load on the card, I would get a black screen (see my earlier post for all of the details.)  Other people were reporting black screens on boot.  Some were seeing various random black screen lockups that seemed to have no direct connection to anything.

There were a bunch of theories as to what was causing these; overclocks that were too much, bad GDDR memory, heat issues including bad thermal paste, PSU's that weren't capable of running the card.  Sorting through the various theories was made even more difficult because some of the people reporting the issues were using various modded bios versions (some for bitcoin mining, some for modding the R9 290 to make it perform like the R9 290X) that may have been causing the issues as well.  In my case, one theory that seemed possible was that the DVI ports did not have a good connection, and was causing the GPU to lock up.  The person with this theory was having the exact same crash that I was.

Regardless of the reason, the first board I received would not run any game for more than a few seconds without black screening and locking up.  With my new board I looped the Metro: Last Light benchmark for about 10 minutes with no problems, as well as running through all of the benchmarks I mentioned without issue.

I will have some time over the next few days to really put it through some gaming to make sure there are no other problems, but so far so good.

Thanks for the info. Good troubleshooting and burn-in testing now. Maybe try OCCT PS Test?

Ya, single chip is usually more trouble-free than CFx unless you are running Eyefinity and need it.

30-60% better 1080p FPS than 5970 (which is almost 2 of my 5870s) .... wow  

Asus is usually pretty good, but it could have been a fluke. I also see it's a OC model. I'm generally against OC anything. Running anything beyond it's designed speed it usually a bad idea because it's really hard on the electronics (or car engine or whatever).

Thanks for the info. Good troubleshooting and burn-in testing now. Maybe try OCCT PS Test?   Ya, single chip is usually more trouble-free than CFx unless you are running Eyefinity and need it.   30-60% better 1080p FPS than 5970 (which is almost 2 of my 5870s) .... wow     Asus is usually pretty good, but it could have been a fluke. I also see it's a OC model. I'm generally against OC anything. Running anything beyond it's designed speed it usually a bad idea because it's really hard on the electronics (or car engine or whatever).