MSI Caught Overvolting GTX 660 Ti, 670 Power Edition Cards?

[Nelly]

What do you make of this... If true, does this effect the GTX 680 Lightning?

If a company changes its product in the middle of a production run after an error becomes public, you know something’s fishy. If Nvidia steps in as well, that’s a good indication there’s something serious going on. Here’s what happened at MSI.

Our colleagues over at Tom’s Hardware Germany seem to have caught MSI red handed, overvolting the GPUs on the GTX 660 Ti and GTX 670 Ti Power Edition boards (Google Translate) to achieve a higher and longer lasting GPU boost state by basically circumventing the PWM controller. In other words, MSI was cheating. Perhaps no one would ever have known if it hadn’t been for one side effect. The increased voltage can cause the system to refuse to POST.

Full Story here >> http://www.tomshardware.co.uk/MSI-GTX-660-670-overvolting-PowerEdition,news-40278.html

The full article on Tom’s Hardware Germany (Google Translate) covers the topic in more depth >> http://www.tomshardware.de/MSI-GTX660Ti-Overvolting,testberichte-241108.html

 

[railven]

This factoring the EVGA news, does this mean these cards may be pulled from circulation?

Glad I only buy reference.

 

[Mastercommander]

Had been posted but on a non descript thread, thanks for repost.

Since some of the German articles are the same, here I was thinking they just had a low opinion of American readers and decided to skimp on these technical details

 

[blackened23]

Shame on MSI for providing a card better than the competition for the same price.

Seems like a ridiculous way to slam MSI, I still consider MSI cards the best. The lightning 680 is the best 680, and the PE 670 is the best 670 on the market, period.

 

[Nelly]

I've got the MSI Z77 MPower and an MSI GTX 680 Lightning, it's annoying that Nvidia are forcing manufacturers to remove voltage options, but then this comes to light...

I did read this though on the MSI Forums >> http://forum-en.msi.com/index.php?topic=162140.0

1.) I'm the author of the German article published on Tom's Hardware.
2.) The article has a real background and there is no sensationalism, the test is based on multiple feedback from our readers after the GTX 660 Ti roundup.
3.) One lab, one of our readers and I have independently reached the same results and analysis of errors in the wiring.
4.) I have informed MSI first and waited for a statement before we decided to publish the article.

The MSI headquarter has responded officially:

... We are currently not aware of any other vendor providing this same level of performance and we’ve worked with NVIDIA to ensure that new production models will limit this free overclock boost you currently get. Our new production models with normal GPU Boost function will be on sale next month.


Just trying to read between the lines

A little part in English from our news - but: "Sorry, you are not allowed to post external links. "
Nice forum

Personal note:
This heavy overvoltage (I'm waiting for a response from Richtek) is a dirty cheat. I took a Gigabyte 660 Ti (with the same combination of TL431 and RT8802A) and I have separated the anode of the TL431 electrically in the same way. What I got was the same result - instable, high VDD, a lot of boost and some problems to start my computer. We have recommended the Power Edition in our roundup verbally and I feel cheated as a reviewer as well as the readers. Sorry for my poor English but I try to improve it

Additional:
Two other manufacturers have already had the MSI card in their laboratories and at least one has officially protested at Nvidia. That means: it would have been a matter of time when this news would become public. But without technical background. I leave each reader to decide what would have been better.

 

[MarkLuvsCS]

They should totally go the way of EVGA and sell a special $100 part just to use that feature. It obviously makes more sense .

I should have my MSI 670 PE by weds so yayay me.

 

[SickBeast]

Why would more voltage cause the card to fail to post? That makes no sense.

nVidia has been very silly in locking down their cards so heavily this round. It has given AMD a big advantage in overclocking. If it weren't for this we would probably see more people going with nVidia.

 

[railven]

Why would more voltage cause the card to fail to post? That makes no sense.

nVidia has been very silly in locking down their cards so heavily this round. It has given AMD a big advantage in overclocking. If it weren't for this we would probably see more people going with nVidia.

From what I read and understood at the Tom's article, the card is requesting almost double the volts required and in certain situations, the PSU isn't expecting to deliver that much increased volts thus the card doesn't power up.

At least that is how I understood it, I'm not an electrician/engineer.

 

[blackened23]

There is absolutely no way any of these request double the volts, if the article says that it is pure sensationalism. I've used a DMM with my MSI lightning 680s quite a bit and on BIOS 1 it always stays within spec, voltage wise. LN2 BIOS allows for higher voltage obviously. Let's be rational: A card requesting "double volts" would instantly destroy itself; even .1V higher causes a substantial increase in heat output.. 1.5V is considered an area only for LN2 benchers, any card attempting that on air is not going to last. A card using 2V for the core? It would instantly die unless it was under heavy duty LN2, and even then it could still die.

If the article is suggesting a card requests 9.3V or double volts? Yeah right.

 

[SirPauly]

Interesting!

 

[3DVagabond]

What? Someone is using the boost function to cheat and pad the benchmark results in their favor! Who would have seen this coming?

 

[RussianSensation]

Blaming MSI for "cheating" by overvolting cards misses the far bigger picture here: NV stopped voltage control overclocking on the entire GTX600 line.

If NV stops voltage control from now on, that has far more serious long-term effects on the consumer, especially for those trying to save some $ and getting a cheaper card and overclocking it.

It's pretty interesting how NV slowly is taking enthusiast features right out of their cards, while AMD instead introduced dual-BIOS switches and sent out Boost/Ghz edition BIOS to the public and tells them it's fine to flash their cards. Talk about a completely 180* from the days when Albratron Ti 4200 could hit 4600 performance speeds with its infamous 3.3ns memory and cards like EVGA GTX460 FTW, MSI HAWK that could hit 925-950mhz.

There is more than meets the eye here and MSI PE/Lightning cards are just casualities of the bigger picture -- NV to disclosing info that something else is behind the scenes. For years and years NV never complained about overclocking and let AIBs have free reign more or less, but suddenly this all changed this generation. Why?

 

[SickBeast]

What I don't understand is why nVidia has done it. Surely they can implement features to prevent people from killing their cards with too many volts. Throttle the cards at 85C for all I care. I will use better cooling so I can get better performance, and I wouldn't want my GPU that hot and loud to begin with.

To me, the only true reason they have done it is to prevent people from moving up a performance class for free. This would work if they had a monopoly but with AMD offering cards like the 7850 and the 7950 it's really hard to say no to a 40%+ overclock on a GPU which is historically unheard of.

 

[3DVagabond]

Blaming MSI for "cheating" by overvolting cards misses the far bigger picture here: NV stopped voltage control overclocking on the entire GTX600 line.

If NV stops voltage control from now on, that has far more serious long-term effects on the consumer, especially for those trying to save some $ and getting a cheaper card and overclocking it.

It's pretty interesting how NV slowly is taking enthusiast features right out of their cards, while AMD instead introduced dual-BIOS switches and sent out Boost/Ghz edition BIOS to the public and tells them it's fine to flash their cards. Talk about a completely 180* from the days when Albratron Ti 4200 could hit 4600 speeds with its infamous 3.3ns memory and cards like EVGA GTX460 FTW, MSI HAWK that could hit 925-950mhz.

There is more than meets the eye here and MSI PE/Lightning cards are just casualities of the bigger picture -- NV to disclosing info that something else is behind the scenes. For years and years NV never complained about overclocking and let AIBs have free reign more or less, but suddenly this all changed this generation. Why?

But nVidia locking voltage isn't news. Everyone who's a regular here should know that already.

 

[RussianSensation]

Market positioning is probably a big factor. GTX480 and 580 had more VRAM and larger differences in specs and performance vs. 470/570 than 680 has over the 670. Maybe NV planned to release a re-fresh of 680 with 1250-1300mhz clocks later but abandoned those plans where 7970GE didn't beat 680 enough to warrant such an SKU?

The RMA could be a factor but if this was a factor, how come for 10+ years it wasn't implemented?

But nVidia locking voltage isn't news. Everyone who's a regular here should know that already.

I know that but why won't they allow AIBs to create custom cards that can take the voltage increase safely? I guess the article mentions there are problems in terms of long-term life expectancy and issues with GPUs posting/black screens. At the same time that still doesn't explain why a manual overclocking option in MSI AB doesn't work on GTX600. Overclocking can cause instability anyway and the user is aware of this. This time NV simply removed that option. I guess GK104 was really pushed near the limits and maybe this is why there is not much more headroom left for the average GK104 chip to overvolt/GPU boost like the MSI PE cards do. MSI would have gotten away with this if their cards were 100% bullet-proof stable with this overvolt.

 

[SickBeast]

The "why" is the big unknown here. It will be interesting to see what they do in the next round. If they lock down voltage again I will go with AMD out of principle. I refuse to support a company that hinders enthusiasts. I am an enthusiast to the core and this stuff really bothers me. I try to avoid Intel for the same reason.

 

[tviceman]

Shame on MSI for providing a card better than the competition for the same price.

Seems like a ridiculous way to slam MSI, I still consider MSI cards the best. The lightning 680 is the best 680, and the PE 670 is the best 670 on the market, period.

Agreed. Nvidia is cracking down hard on overvolting. It's either because they know GK104 is bandwidth strapped, and gets very little additional benefit from over-volted overclocking (unless the vram is clocked to the moon), they are genuinely trying to put a stop to overvolt overclocking for cost/warranty/RMA reasons, or they don't want users extracting 30+% more performance out of their GPU's which they feel was eating into upgrade cycles.

I think it's probably all of the above.

 

[hyrule4927]

Glad I only buy reference.

Don't hear that very often . . .

 

[SickBeast]

Agreed. Nvidia is cracking down hard on overvolting. It's either because they know GK104 is bandwidth strapped, and gets very little additional benefit from over-volted overclocking (unless the vram is clocked to the moon), they are genuinely trying to put a stop to overvolt overclocking for cost/warranty/RMA reasons, or they don't want users extracting 30+% more performance out of their GPU's which they feel was eating into upgrade cycles.

I think it's probably all of the above.

Yes that's an excellent explanation. I was forgetting about the fact that Kepler is limited by vram frequency.

 

[3DVagabond]

I think they are just trying to lock in the performance brackets. The 670 has to be eating the 680's lunch sales wise. That's $100 every time someone buys a 670 instead of a 680. It will also allow them to release more SKU's. Look at Intel with their frequency locked server CPU's. They can release many more models over a much larger price range than they can with their consumer CPU's that can be O/C'd.

 

[Mastercommander]

Lol some of you guys should read the article instead of each other.

Moving on, the language tomshardware uses might be a tad strong. That said, I'd be mad if my card refused to POST.

I find fascinating that it was the Germans who elucidated it. Not surprising

 

[Sharchaster]

Agreed. Nvidia is cracking down hard on overvolting. It's either because they know GK104 is bandwidth strapped, and gets very little additional benefit from over-volted overclocking (unless the vram is clocked to the moon), they are genuinely trying to put a stop to overvolt overclocking for cost/warranty/RMA reasons, or they don't want users extracting 30+% more performance out of their GPU's which they feel was eating into upgrade cycles.

I think it's probably all of the above.

Excellent explanation, that's what I'm thinking so far. Lucky for me to bought PE card much faster. nvidia is like a child IMO to prevent their ENTHUSIAST users to got enjoyable overclocking experience.

Nice for MSI to take a risk for offering their users to got enjoyable overclocking experience, which other brand can't do it.

 

[NickelPlate]

Pardon my ignorance but can someone please fill me in on this removal of voltage options that NV has apparently stopped or is going to stop? I can adjust voltages on my GTX670 using precisionx. Is that not the same thing or is it gimped some how?

 

[blackened23]

Pardon my ignorance but can someone please fill me in on this removal of voltage options that NV has apparently stopped or is going to stop? I can adjust voltages on my GTX670 using precisionx. Is that not the same thing or is it gimped some how?

The slider in precision X doesn't adjust max voltage. It adjusts vdroop only, meaning that it adjusts your idle state voltages.

1175mV is the stock voltage for load and you can't adjust that

 

[exar333]

Blaming MSI for "cheating" by overvolting cards misses the far bigger picture here: NV stopped voltage control overclocking on the entire GTX600 line.

If NV stops voltage control from now on, that has far more serious long-term effects on the consumer, especially for those trying to save some $ and getting a cheaper card and overclocking it.

It's pretty interesting how NV slowly is taking enthusiast features right out of their cards, while AMD instead introduced dual-BIOS switches and sent out Boost/Ghz edition BIOS to the public and tells them it's fine to flash their cards. Talk about a completely 180* from the days when Albratron Ti 4200 could hit 4600 performance speeds with its infamous 3.3ns memory and cards like EVGA GTX460 FTW, MSI HAWK that could hit 925-950mhz.

There is more than meets the eye here and MSI PE/Lightning cards are just casualities of the bigger picture -- NV to disclosing info that something else is behind the scenes. For years and years NV never complained about overclocking and let AIBs have free reign more or less, but suddenly this all changed this generation. Why?

While I agree with your post, I don't think this has any relevance with the original story. The real issue here is that MSI was allegedly modifying these cards in such a way that they would actually not boot-up at all. That is a relatively serious issue, and I could see how NV would take this as a bad effect on their product brand. I don't like OC limiting either, but getting a product right out of a box that is possibly defective because of what the AIB did, is a whole other situation.