GTX Titan Owners

[radrok]

Just saw some truly horrific frame time benchmarks of HD 7970 CF. In few minutes after that I ordered EVGA GTX TITAN SC. Never ever again SLI or CrossFire

http://www.pcper.com/reviews/Graphi...ils-Capture-based-Graphics-Performance-Test-5

Well done sir, still SLI isn't that bad if you want to go multi GPU.

I'll post some SLI results in a few days, more like I'll give my experience about it.

 

[radrok]

I'm having delights by rendering in Octane with dual Titan, scaringly fast.

 

[tastegw]

Got mine installed earlier last week

Evga sc edition, ordered a backplate for it also, can't wait to install it

 

[xorbe]

Wow I'm actually glad I went with Gainward, it is not probably related to what I'm going to say though.

This card boosts at 1100MHz WITHOUT touching any slider, wow what a good chip I've got.

The Zotac one boosts at 998 MHz.

That is a good card if it's actually stable. I don't think I've seen anyone claim 1100 speed at stock settings. Mine settles at 967. :shadedshu With naennon bios, it runs 1097 @ 1.187v fine with everything I've thrown at it including FC3 and 3DMark11. I've got a couple bins of headroom for stability.

 

[radrok]

Yes it's stable but it won't go over 1202 MHz on stock BIOS, although it can run valley on 1250 MHz.

A soon as I have some time to toy with it I'll try a higher voltage higher power target BIOS.

The Zotac one is not as good and enabling SLI seems to reduce my max boost on the fastest card

 

[the54thvoid]

1202 is nice mate... I'm jealous.

I have mine down at 1097 on stock volts, +200 memory (very modest). But it's solid at that speed and 200+MHz above advertised boost speed. My card defaulted to 993 boost without overclocking. That seems a very common figure.

Yours is bloody marvelous Radrok.

I've yet to try naennons bios but i may give it a whizz. My card runs as high as 1189 using EVGA Precision but is unstable. If i try anything higher it immediately has a driver crash and resets to stock. I think my low asic means i require the extra voltage and power headroom that naennons would offer.
But if that means 100MHz extra for only maybe a few extra frames it's not really worth it. It's bloody fast at 1097, shouldn't complain.

Oh, Radrok, you ought to run your card through 3DMark11. Would love to see 1202 scoring high on that.

 

[radrok]

I will, did you OC the memory too? It seems to give very good gains but it becomes hot as hell, how cheap of nvidia not to include a backplate lol

 

[the54thvoid]

I've only overclocked the memory by +200 (6412MHz). Not really tried going way high as i don't like the idea of the memory chips getting too hot.

I've actually just tried a self modded BIOS with a power limit at 300watts and voltage in 3D (P0 and P02) at 1.21v. I managed to get into Valley with 1202MHz core and thought I'd run the benchmark. I got half way through and it crashed

It appears my low asic card is high leakage and needs more powah (specifically voltage). I'm not willing to try anything more though. It crashed out on stock bios at 1.2 volts at 1176/1189 so it definitely is a voltage weakness.

I'll try and find my highest stable clock on the most efficient volts (above 1100MHz). Actually crashed at 1097MHz on 1.6v (stock) after 2 1/2 hours of BF3 the other night. So I'll spend some time at 1110MHz at 1.18v and see how that goes. I'm perfectly happy with 1100 ish core tbh. I'll move the memory up bit by bit and see how far and not too hot it can get.

Boost 2.0 is not an issue to me under water but the power limiter and low voltage is

 

[radrok]

Boost 2.0 is not an issue to me under water but the power limiter and low voltage is

Imagine what kind of monster we'd have if it was produced as a Lightning variant.

Titan + Lightning VRM....... one can only dream

 

[tastegw]

stock boosts on my card topped out at 1058, will OC later on when everything is all squared up.

 

[HammerON]

I wonder how these cards do at crunching or folding...

 

[bpgt64]

I am trying to get mine online and folding, but I didn't see a Window's 8 FAH gui less installer.

 

[HammerON]

Saw this:
http://foldingforum.org/viewtopic.php?f=16&t=21834

But I am not sure....
Still using Windows 7

Ask the folding folks here:
http://www.techpowerup.com/forums/showthread.php?t=182493&page=6

 

[thebluebumblebee]

I am trying to get mine online and folding, but I didn't see a Window's 8 FAH gui less installer.

Win7/Win8 uses the same client. There's really only the GUI client now.

 

[johnerz]

Just download and run the install it will recognise the Titan (if you have version 314 drivers) DO NOT set Advanced or Beat flags, it shouls work right of the bat

This may help

http://foldingforum.org/viewtopic.php?f=38&t=23944&start=15

 

[radrok]

Oh, Radrok, you ought to run your card through 3DMark11. Would love to see 1202 scoring high on that.

http://www.3dmark.com/3dm11/6422100

max boost was 1228 MHz but it throttles back because of the power limit

 

[the54thvoid]

Is that on stock BIOS and what oc tool you using? My card only boosts to 993 on stock settings but I've been using 1124MHz core and 6500MHz memory with 1.18v. That saw me stable last night through 2 1/2 hours of BF3.

I'm tempted to try the engineering BIOS that Metal Racer is using to bench with but it looks very odd when loaded onto KBT.

My standard BIOS is on the left, the engineering sample, handily an ASUS one, on the right. The engineering sample is locked completely when opened in KBT so you can't see the voltages etc. Metal Racer has used it but I'm a wee bit cautious. The card is under water though so temps are not a concern - just don't like the unknown...

I'd just love to run my car at 1200MHz for gaming - Can't explain why - just do. I'm sure 1124MHz to 1200MHz is irrelevant in fps numbers though. And I've not really pushed the memory.... Maybe someone needs to tell me to sit down and calm down!

 

[radrok]

Yes that's on stock BIOS, I use a rivatuner derivative should be Precision X or Afterburner, they're the same tho so np.

I haven't had the time to try and flash that BIOS, how is it on idle clocks?
Keeps the voltage pegged at 1.2v?

 

[the54thvoid]

I've not tried it either. I've modded my own bios and it gave me a couple of extra clock steppings but my card leaks voltage (based on oc results and upping power limits and voltages). My own bios modded went to 1.21v and it downclocked for 2D but if these engineering bios's are stable and allow a higher power and voltage draw (while still being safe for 24/7) I'd be interested to try.

My stock bios oc limit is 1176MHz @ 106% power draw and 1.2v - launches but can crash during Unigine Valley
Using modded power (300w max) it still acts as above.
If I mod my voltage in bios to 1.21v it launches Valley at 1202MHz but still crashes. At 1.2v my card will not even launch at 1202 - instant driver crash.

I believe 1.21v is the cards maximum and hard mods (not my area at all - nor am i interested) are required to go higher. Maybe I should accept the core and play for memory... Should add I am still very happy with card. Stable 1124Mhz is still 127MHz above its max stock boost and still just under 300MHz (or 34%) higher than base clock

 

[the54thvoid]

Well, I got round to flashing 'naennons' bios this morning. Had to reinstall my drivers afterwards but it all went well. I had modded the bios before installing as it's set to boost to 1398. I had pegged it nice and low at 1202.
There is something mystical about the bios itself (I believe it was hex edited properly, not just in KBT). My own modded bios run at 1202 with 1.21v @ 300watts failed but with this running at 1.212v and just 100% power (260watts) it boosted to 1202 and ran the valley bench without issue.

The irony is, the benchmark score at 1202 core, 6008 memory was 70.9 (Extreme HD preset).
When i re-installed the stock bios and ran clocks at 1124core 6500 memory @1.18v, the score was also 70.9.

Questions I'd love to have definitively answered are...

What is safe voltage? My OC runs 3D at 1.18v. If 1.212v is the cards hard limit, is it safe to game on it?
Also, if i incrementally increase memory clocks, will it artifact long before any thermal damage is done?

Obviously I'm on water and backplate.

 

[xorbe]

I never got a clear answer: either he edited another hidden value in the rom, or the nvidia vendor code vs evga/asus vendor code has a strange engineering sample effect.

 

[radrok]

Well, I got round to flashing 'naennons' bios this morning. Had to reinstall my drivers afterwards but it all went well. I had modded the bios before installing as it's set to boost to 1398. I had pegged it nice and low at 1202.
There is something mystical about the bios itself (I believe it was hex edited properly, not just in KBT). My own modded bios run at 1202 with 1.21v @ 300watts failed but with this running at 1.212v and just 100% power (260watts) it boosted to 1202 and ran the valley bench without issue.

The irony is, the benchmark score at 1202 core, 6008 memory was 70.9 (Extreme HD preset).
When i re-installed the stock bios and ran clocks at 1124core 6500 memory @1.18v, the score was also 70.9.

Questions I'd love to have definitively answered are...

What is safe voltage? My OC runs 3D at 1.18v. If 1.212v is the cards hard limit, is it safe to game on it?
Also, if i incrementally increase memory clocks, will it artifact long before any thermal damage is done?

Obviously I'm on water and backplate.

How did the card idle? Was voltage at 1.212v even on 2d clocks?

I have to find some time to play with the engineering BIOS. Such a shame these cards top off a 30c and there is no voltage to toy with

 

[the54thvoid]

How did the card idle? Was voltage at 1.212v even on 2d clocks?

I have to find some time to play with the engineering BIOS. Such a shame these cards top off a 30c and there is no voltage to toy with

No, 2D clocks were below 0.9v. They just stay locked at 1.212v during gaming instead of varying with load. But to be fair, my OC stays at 1.18v when gaming, with no down volting unless i exit the scene.

If i didn't have to reinstall the driver each time I'd play with it more - try a higher boost than 1202 but to be honest, I know now the card can run above 1200MHz. It just needs special help. So the silicon must be good enough, just the stock BIOS and Nvidia's uber caution.

I hope Nvidia are looking at drivers because you feel that they have so much potential but are being held back by a lower than necessary power limit and voltage. Not that 15k+ on 3DMarkk11 is 'poor'.

 

[radrok]

No, 2D clocks were below 0.9v. They just stay locked at 1.212v during gaming instead of varying with load. But to be fair, my OC stays at 1.18v when gaming, with no down volting unless i exit the scene.

If i didn't have to reinstall the driver each time I'd play with it more - try a higher boost than 1202 but to be honest, I know now the card can run above 1200MHz. It just needs special help. So the silicon must be good enough, just the stock BIOS and Nvidia's uber caution.

I hope Nvidia are looking at drivers because you feel that they have so much potential but are being held back by a lower than necessary power limit and voltage. Not that 15k+ on 3DMarkk11 is 'poor'.

There are some guys that used the power sense short (it's an hardware modification tho) that achieved 17k++ with just 1.3v

I honestly feel that nVidia should have been more forgiving with limits and they surely shouldn't have saved a few dollars on the VRM section.

If AIBs can fit the additional custom vrm cost in just a 30-40 eur increase over reference then nVidia has no excuses.

 

[radrok]

First non render shots of the Aquacomputer XSC Backplate.

Gorgeous, thank goodness I waited for Aquacomputer blocks.

Gonna order them as soon as they are available.