79XX Dont Run

XP support was dropped from the latest SDK release

Means those on XP (like myself) will need to upgrade Windows to use the 79xx series GPUs, I presume?

In the meantime the HD69xx and HD58xx are OK as XP graphics drivers are available?

---

Go away, I was asleep

XP support was dropped from the latest SDK release

One way or the other Microsoft will get it's Blood Money from the Masses ... :P

What puzzles me is then why even have XP 32 & 64 Bit Drivers on the 7970 CD & it shows installing the SDK Drivers too if their not going to work ???

---

STE\/E

Means those on XP (like myself) will need to upgrade Windows to use the 79xx series GPUs, I presume?

For 79XX highly likely - lets wiat for the first big driver update, but I'll be amazed if XP is there for 79XX if its dropped from the SDK.

There will always be the drivers already deployed, but with MS only looking after XP on extended Hotfix/Security basis only no Service Packs until dropped totally 2014, the operating environment for those who stay on XP is going to degrade slowly but surely. - its inevitable.

From what Matt has said, it would appear AMD has also kicked XP into touch if its dropped from the latest SDK. Life will not suddenly grind to a halt in XP land, but the evil day is approaching when Win7 cometh like it of not :) Certainly if its dropped from the SDK, that will mean future builds of (whatever) will not include XP, and XP users will fall behind fairly quickly, months not years.

If I were you I'd bite the bullet now, and plan out a migration route, before reality forces it on you in a hurry.

Regards
Zy

May have to reluctantly go to Win 7 or just forget about Upgrading to the 7970's ... :/

If you have to buy several licenses that'd be painful. If you can *try* one for free, e.g. over MSDNAA, that might be a less painful start. If you look at it open minded (e.g. you do not want everything to be exactly as it has always been) it's actually amazing how much things have improved over the years.

I used to hate how MS always tried to think for me and do everything automatically.. and never got it quite right. With Win 7 they're finally getting there, I can lay back and mostly just let it do its thing, without me having to worry about stuff. I tried it when it came out and never looked back.

If it's just for a cruncher you could consider getting used Vista licenses. The graphics driver model is the same as for Win 7, so will last a lot longer than XP. For actual work it's much less pleasant than Win 7, though.

MrS

---

Scanning for our furry friends since Jan 2002

If I were you I'd bite the bullet now, and plan out a migration route, before reality forces it on you in a hurry.

Regards
Zy

Already have plans in the works, Free Copies of Win 7 Professional courtesy of a friend ... ;) ... Just need to find the time to convert all the Box's or at least a few to be able to run the 7970's properly ...

---

STE\/E

I reckon you'll end up liking it ..... as long as you stop thinking of reasons not to like it, and start looking at reasons why you should like it :)

You'll certainly find management of those copious boxes of yours easier.

Free ..... friend ...... hmmmm tusk, tusk, wash your bank account out with soap and water rofl :)

Regards
Zy

I have Win 7 installed on my Laptop, Hate it really ...

---

STE\/E

I've got 2x7970 in crossfire...

any particular reason you're crunching with the cards in X-fire? seeing as how neither X-fire nor SLI scales perfectly, 2 AMD/nVidia GPUs in X-fire/SLI will never produce twice the performance of one of those GPUs. 2 AMD cards (or nVidia cards) not in X-fire (or SLI) on the other hand will have twice the compute power. i'm assuming you game part of the time and don't want to hassle with regularly enabling and disabling X-fire, or have some other good reason for running those GPUs in X-fire even though its generally counterproductive to GPGPU computing?

---

They are not running crossfire now - was just trying them out. I agree, same result for my 2x5970s in the other box, best not to crossfire in BOINCLand if possible

I have Win 7 installed on my Laptop, Hate it really ...

It kind of grows on you, I have it on all but two of my pc's, one is still XP and the other is Windows Home Server 2011. I have both 32 and 64 bit versions and even put a copy on my Netbook, it is better than XP or Vista IMO.

Matt,

There seems to be an issue with the OpenCL app and 7970s, not 100% certain yet, but the symptoms this end are consistent. I get the cards running after a reboot, and setting Afterburner (voltage unlocked, Kernal Mode, supply constant voltage), cards set at (at the moment) 1.218v / 1185 GPU / 1375 Memory / fan 53. The cards will run for about 2-3 hours, then the invalids start. Not many as such, but constant. If I reboot and reset, same pattern.

I have tried different voltages (went up to 1.225v which is a crazy level for the settings on the card I have been using), and GPU down to 1170 up to 1230. Whatever the mix of GPU speed and voltage, still get the same pattern. I'm not going above 1.225v or 1230GPU, not happening (!).

I am running Prime Grid PPS LLR challenge on that PC at present, using four cores of the 1090T, leaving two cores to service the 7970s. Its possible that one core per GPU is needed, and there is a clash only having two cores supporting the two 7970s. These things are a beast when o/c and its possible two cores are being drowned. Hard to say as CPU core utilisation whilst still well down will not really show a clash between the CPU app and GPU app.

The Prime Grid Challenge ends at 1800UTC, and at that point I will switch the CPU to WCG Energy Phase 2 on two cores, and leave four cores for the 7970s to see if makes a difference - didnt want to do an extended test on four cores/7970 whilst the PG Challenge was running.

Whatever it is causing it, its going to be hard to spot from this end I suspect. See what happens when I switch over to WCG and leave additional Cores to the 7970s.

Regards
Zy

Interesting .... its not software .... its a hardware limiation. I took two cores away from the PG Challenge as that will not affect my Team position, and currently running at 1.206v / 1180 GPU / 1375 Memory / 53 Fan. So far (get back Murphy) no invalids for over an hour. Dont yet know if its using four cores that did it, or speed reduction to 1180 - which was just 5 less than before. I have a feeling its the number of cores available.

I'll give it a while then turn it back up to 1185, see what happens. However, getting near the limits for the card / app / PC hardware, just need to find out now if its GPU speed or my hardware that is the limiter.

In some ways its similar to a Guru3D test they did on different CPUs and 7970s, they were surprising results .... worth a look if you into this end of hardware/crunching.

http://www.guru3d.com/article/radeon-hd-7970-cpu-scaling-performance-review/7

Anyone else with a 7970 and ventured into overvolting, would be interested to hear what you found, and the hardware spec of the PC.

EDIT: Posted the above and .... Murphy heard me, an invalid 2 mins later rofl:) I'll leave it on four CPU cores for the 7970s and 1180 GPU/1375 Memory for a few hours now, see what happens.

Regards
Zy

Three hours now .... no invalids. Looks like running on four cores was the answer. I went up to 1.206v / 1240 GPU / 1375 / fan 53, and thats me ..... noooo further, going higher is for the brave /suicidal. It does however verify the extreme overclock I've seen reported at 1260.

I found minimal improvement at 1230 & 1240, went faster for sure, but not really worth the effort and grief running 7x24. The WU has settled at a solid 44 secs a go, remarkable considering its a NVidia OpenCL app. Temperature @1220 is 65 on one card, 67 on the other - there's a CPU air cooler just above one so that would account for slight difference with warm air coming out the CPU cooler. Temperatures are steady, only slight - veeery slight - blips showing on Afterburner as WUs load / unload.

Someone running them on a Sandybridge will go faster I reckon, but not by much, it'll be a function of GPU clock speed, and how good case heating is. Certainly going past 1230 to 1240, performance improvements were starting to flatline, and temperatures rose a notch .... so at that point that was me, backed of to 1220, and will happily run that7x24 with WUs running @44secs. This card definitely still has more legs in it if its taken via water cooling, I stopped at 1240, voltage 1.206v, the card will take 1.3v ..... so on water .... well who knows., but me a happy teddy @1220 on air 7x24, that'll do me :)

Good session, pleased with that :)

Regards
Zy

Three hours now .... no invalids. Looks like running on four cores was the answer.

this may seem a bit nit-picky, but what do you make of the 2 invalids caused by Murphy in the above post? you had already relegated PG to 2 CPU cores, and had 4 spare CPU cores available for the MW@H GPU tasks when those invalids happened...

---

Three hours now .... no invalids. Looks like running on four cores was the answer.

this may seem a bit nit-picky, but what do you make of the 2 invalids caused by Murphy in the above post? you had already relegated PG to 2 CPU cores, and had 4 spare CPU cores available for the MW@H GPU tasks when those invalids happened...

No problem ... sanity checks always a good thing, I'm searching for a stable setup that I can sleep at night with ... but still make best use of the cards.

If you look at the valids/invalids there was an invalid at 1926, I then switched to four cpu cores. Since then I have only had one invalid, at 2051hrs, and I have a suspicion I caused that - was messing with other non-BOINC things. At that time I was down at 1200 GPU. No invalids since despite going to 1240. So I reckon its pretty solid.

I'm going to let it run overnight at 1.206v / 1220 / 1375, then tomorrow step down the voltage a bit. Card default voltage is 1.174v, so its not a high voltage change, frankly its low in the scheme of things, but will try tomorrow to get it down a bit once I'm 110% certain of 1220 GPU on an overnight run.

Regards
Zy

grrrr ..... 2 errors and an invalid :)

Guess its not quite there yet. I'll leave it at 1.206v / 1220 overnight, see what its like in the morning.

Regards
Zy

sounds good...looking forward to your next update.

on a side note, i've never been able to eliminate invalids altogether on my HD 5870, and it never occurred to me that it might be a memory clock (or more appropriately a memory bandwidth issue) until i read through this thread and saw your testing methodology. i run the core clock at the stock 850mhz and the VRAM at half the stock frequency, so 600mhz. my GPU goes through well over 1000 MW@H tasks per day, and only yields a handful of invalids...i can't recall exact numbers, but if i round up the number of invalids to 10 per day and round down the number of tasks completed to 1000 per day, that only equates to a 1% invalid rate, and in reality i think its probably closer to 0.5%...not bad. yet i wonder if i should be concerned about this and continue to troubleshoot it like i have, or just let it go.

i increased the memory clock from 600mhz to 700mhz (since underclocking the VRAM too low can also yield adverse affects just like overclocking the VRAM too much can) and let it run for several hours, and it actually increased the number of invalids i usually get. i then tried VRAM at 800mhz, and the number of invalids came back down to about as many as i was getting when VRAM was set to 600mhz. perhaps if i continue to raise the VRAM frequency closer to the stock 1200mhz in 100mhz increments might yield positive results, but for now i've stopped testing...

---

Interesting .... its not software .... its a hardware limiation.

All of your invalids look like the entire output is the letter 'E', so it's probably the truncated stderr problem

Interesting .... its not software .... its a hardware limiation.

All of your invalids look like the entire output is the letter 'E', so it's probably the truncated stderr problem

Thanks Matt, thats useful to know.

It blue screened after my last post (down to me that one, being mean on voltage I think), its back up and been running a while with no invalids. So I'll "restart" the invalid count from this point and see how she goes.

Its running now at 1.218v / 1220 GPU / 1375 / 53 fan.

If it bombs out again, I'll step up volts another level, but going no further, from that point I'll start declocking below 1220. I still think it will run ok with a sensible voltage at 1220, just need to go there one step at a time.

Lets see if Mr Murphy presents me a Blue Screen for my pains when I wake up in a few hours :)

Regards
Zy