Updated firmware for EVGA all 2gb GTX 670 cards

[blackened23]

This is a very popular card so I figured this would come in handy:

Copy pasted from http://www.evga.com/forums/tm.aspx?m=1703680

EVGA GeForce GTX 670 Firmware Update v80.04.31

Q. What are the changes in this new firmware?
Improved stability
Improved compatibility with DELL XPS 8300 systems

Q. Do I need to reapply this update every time I update a driver or reinstall the O.S.?
A. This update is attached to the card, no matter where it goes. If you update the driver, the EVGA firmware update is still applied.

Q. How can I tell what version I firmware I currently have?
A. There are multiple ways to do this, one way is to right click on desktop and go to NVIDIA control panel. Click on Help, then System Information. Scroll to the bottom of the Details window and it will show your "Video BIOS version". It will be listed as EXAMPLE: 80.04.31.00.72. The numbers in bold identify the version number.

Q. I just bought my card, will it have this updated firmware?
A. Chances are yes, as new firmware is release, new products shipping from EVGA will also have this firmware version applied.


Installation
Download the correct .zip file below for your graphics card below.
Extract the files to a location on disk.
Make sure no programs are running in the background, and double click update.exe.
Press "Y", then ENTER to begin the update.
DO NOT TURN OFF POWER OR RESET DURING THE UPDATE PROCESS!
After update, restart PC.

EVGA GeForce GTX 670 2GB:
v80.04.31 02G-P4-2670-KR - EVGA GeForce GTX 670
v80.04.31 02G-P4-2672-KR - EVGA GeForce GTX 670 SC
v80.04.31 02G-P4-2676-KR - EVGA GeForce GTX 670 FTW LE
v80.04.31 02G-P4-2678-KR - EVGA GeForce GTX 670 FTW

 

[cmdrdredd]

My bet is this is a release from Nvidia that addresses the same things they were addressing on MSI cards, blocking your overclocking. Has anyone used this yet?

This doesn't look very good. Over on evga forums i found this. If it remains true I'm not gonna touch my firmware.

OK, in deed the higher (closer to borderline of stability) overclocks become more stable now... but I noticed at what price... that's all thanks to lower actual TDP under load so it doesn't exceed the Power Target too much (which might have been a problem initially in original BIOS). .... however it manages to keep the TDP at lower levels thanks to more aggressive power throttling on the core in extreme situations.

... so in summary, yes the overclock is stable (firmer now), but might bring slightly less performance because of deeper power throttling in order to get there...

e.g. here on my updated GTX 670 and OC'ed to +145core (1243 actual) & +525mem (3522 actual) I ran 3dmark11 (more stressful than Heaven3.0) with graphs from precision. just look how deeply it undercuts the core clock boost during the Test#1 in 3dm11.



check the graphs on right side of screen, 2nd from top is the actual TDP and the 3rd from top is the Core (should be constant under load). as you note my core OC was throttled down by nearly 150MHz, however it seems that it happened before the TDP spike occurred (GPU BIOS compensating in advance?), so the TDP max levels were fine at a cost of lowered performance.

on original BIOS it never power throttled by that much even at higher clocks, but the TDP was spiking well above 132% too (now no more than 127%).

What I don't really understand is what they mean by stable? Let me put it this way. If you were getting 1250Mhz boost and it held it until 70c and then went down to 1237Mhz boost and staid there the whole time you were playing a game and you never had an error, crash, or artifact. What is not stable about it? I've never seen my memory clock change at all ever, not even when the boost clock is throttling. How is an overclock more stable if it is dropping your clock speed by 150Mhz because of a limiter placed in the firmware/bios? To me that doesn't mean stability at a given clock speed.

 

[thilanliyan]

What is it with these BIOS updates limiting OCing for these cards? There are no mass reports of failures are there?

 

[cmdrdredd]

What is it with these BIOS updates limiting OCing for these cards? There are no mass reports of failures are there?

I haven't seen anything like that. I think Nvidia is going to roll out a new card and don't want the $400 GTX 670 to steal the thunder. So if they can cripple your cards and force them to throttle even more, they can release a GTX 685 for $600 and rip off the elitists who think they need 4 of them.

I asked a few questions over on the EVGA thread asking how this is a good thing for me. I really don't understand what some people are saying with stable. If you don't get errors or crash then you are stable. My cards throttle once at 70c and never reach the second point to throttle. My memory clock doesn't change. My base clock is irrelevant as the cards are boosting past that by quite a bit. So...I'm trying to understand what people are talking about with their responses in the thread mentioning increased stability. I don't think describing a 150Mhz throttle because of tighter TDP limits as "stable" is correct usage of the term.

The only thing I can think of is they had a few too many reports of cards going close to the TDP wall and crashing/freezing. To remedy this they simply force the cards to throttle sooner and reduce the TDP you can actually achieve.

 

[Smoblikat]

The whole concept of these stupi boost clocks baffles me. Why anyone would want to have no control of what their clocks are is dumb, my 470s' overclock like a card should. If I want them to run at 700 core, i set it to 700, I dont need to worry about power this or is my card boosted high enough for that, my cards do what I tell them to.

 

[blackened23]

My bet is this is a release from Nvidia that addresses the same things they were addressing on MSI cards, blocking your overclocking. Has anyone used this yet?

MSI hasn't been forced to release cards with an updated BIOS that locks over voltage, so I assume MSI and nvidia came to agreement. The updated BIOS that they did release simply removed the dust removal technology at startup...(and still allows over voltage and 300% maximum power )

If anything I'm guessing this new BIOS addresses some of the faults that EVGA had with the initial batch of 670 SC cards? I remember hearing of some issues shortly after they shipped.

 

[Plimogz]

[...] if they can cripple your cards and force them to throttle even more, they can release a GTX 685 for $600 and rip off the elitists who think they need 4 of them.

Bait and switch? that would be outrageous.

 

[blackened23]

The whole concept of these stupi boost clocks baffles me. Why anyone would want to have no control of what their clocks are is dumb, my 470s' overclock like a card should. If I want them to run at 700 core, i set it to 700, I dont need to worry about power this or is my card boosted high enough for that, my cards do what I tell them to.

Its a double edged sword, I guess. For the average Joe its a good thing because most people don't bother with OC'ing, and for those people it will benefit. But for users who know what they're doing - stuff like the thermal / overpower throttle are obviously bad things because they hinder maximum overclocks.

 

[cmdrdredd]

The whole concept of these stupi boost clocks baffles me. Why anyone would want to have no control of what their clocks are is dumb, my 470s' overclock like a card should. If I want them to run at 700 core, i set it to 700, I dont need to worry about power this or is my card boosted high enough for that, my cards do what I tell them to.

Yeah I'm totally confused. People are talking about more stable base clock. Base clock doesn't change, boost clock does. If your boosting to ~1250 and throttle at 70c you go down to ~1237. Base clock is irrelevant here because your card is actually running faster than the base clock, similar to Turbo Mode on CPUs. Also, I have never heard of memory clock that varies. My memory clock is always stuck to what I set and doesn't change with the boost clock.

I don't really understand what's supposed to be happening, what is happening, and what people say is happening before and after this update.

 

[Termie]

Thanks for the link, OP.

Assuming that EVGA released these to address stock clock stability (since the release notes say nothing about overclocking), it would make perfect sense that EVGA would want to throttle the cards more often to avoid hitting the power limiter. I assume bouncing off that limiter at full clocks would cause more problems than dropping the clocks preemptively.

As for a "conspiracy theory" to make this card look weaker, well, I don't think that would have much effect, as the benchmarks are already out there for this card, and reviewers, should they bother to rebench the 670 on the release of future cards, probably won't flash their cards.

 

[cmdrdredd]

Bait and switch? that would be outrageous.

Lol...I'm just tossing ideas out there. Something seems way wrong about this whole thing, and the MSI situation as well.

blackened, I had a bad SC card but as far as I know they withdrew them all from retail chanels and replaced them with FTW cards. A buddy of mine told me they sent the couple SC cards they had in stock at a local compusa back to EVGA and got FTW cards in replacement. I don't even see the normal SC version listed on newegg. It seems to have been replaced with a FTW LE version.

I'll keep my eye on the EVGA forum thread for further info. See if anyone answers my questions I posted there. Maybe get a little better understanding of what is going on and why I would want to update.

 

[Silverforce11]

What I don't really understand is what they mean by stable?

Stability could mean max OC stability or long term stability. Guess which ones manufacturers (who have to replace defective cards) care about? Perhaps NV feels its just not safe or smart to push their reference designs with barely enough power circuitry for the stock boost operations.

 

[cmdrdredd]

My guess based on a response on the forum...this is a way for EVGA to avoid RMAs for cards that crash when reaching a specific TDP. Since they are using reference PCBs (some of them GTX 680 PCBs) they cannot handle the higher loads sometimes that heavy overclocks and high power target settings will achieve. To fix all of this for them, they limit TDP a bit and make the card throttle.

This update basically forces the card to clock down even before reaching thermal limits by being more aggressive with the TDP monitoring and throttling.

 

[5.0inaYota]

Yeah I must say I am not a fan of the new firmware, my 670 isn't throttling down due to going over the TDP (seen it go up to 129%) and my overclock is stable, so I am not going to give it a try. I will stick with my 670 the way it is! I have never seen my memory clock change either, don't think that is possible. I still don't understand why they even put an adjustment for the power target if it doesn't stop it from going over what you set it too, if you set it to 122% (max) then it shouldn't be able to go over, but it does.

 

[cmdrdredd]

I think some cards had problems when they went over a certain level. That's all.

 

[5.0inaYota]

The new firmware is out for the 670 4GB cards now too. Yeah I don't think all cards respond the same, granted it is app specific if it will, I have yet to see power throttling during BF3 and my card continually spikes to 129% TDP.

 

[blackened23]

My guess based on a response on the forum...this is a way for EVGA to avoid RMAs for cards that crash when reaching a specific TDP. Since they are using reference PCBs (some of them GTX 680 PCBs) they cannot handle the higher loads sometimes that heavy overclocks and high power target settings will achieve. To fix all of this for them, they limit TDP a bit and make the card throttle.

This update basically forces the card to clock down even before reaching thermal limits by being more aggressive with the TDP monitoring and throttling.

So thats what this firmware does? Wow...lame.

 

[5.0inaYota]

Yeah very, I was hoping for something good.

 

[spinejam]

So thats what this firmware does? Wow...lame.

Well, it certainly doesn't do this:



Uploaded with ImageShack.us





Uploaded with ImageShack.us


http://imageshack.us

 

[blackened23]

I don't get it

 

[chimaxi83]

I don't get it

I think his 1325 overclock

 

[spinejam]

I don't get it

It's probably not going to enhance your overclocking once you flash your card to this new bios. :thumbsdown:

 

[Termie]

It's probably not going to enhance your overclocking once you flash your card to this new bios. :thumbsdown:

Did you use the 1.21v flash on your card?

 

[spinejam]

Did you use the 1.21v flash on your card?

Yes. :thumbsup:

This one: http://ultrashare.net/hosting/fl/584179b68b/DCT670_VMOD_FTWBP-200-225

 

[Grooveriding]

Bait and switch? that would be outrageous.

I doubt it is a prelude to a 685. Nvidia has nowhere to go with the GTX 680. The card's memory is the bottleneck as shown by results of 1300+core clocks not gaining anything performance wise.

680 gains more from memory overclocks than anything else. I guess they could put out a factory clocked 6500 or higher memory card, but just doesn't make sense.

I think GK110 will have to be waited on for them to deliver anything better than the 680. In terms of the actual core of the 680, there is nothing more there to offer.