TechPowerUp Impact: MSI Issues "OC Mode by Default" BIOSes

[mouacyk]

Only if it is as simple as flashing the new BIOS. I have a retail MSI GTX 980 Ti Gaming that becomes unstable when flashed with the review sample BIOS from TPU. That's going from a boost of 1228 to 1279.

No its not a bad flash. It crashes in Witcher 3 and bf4 after an hour or so whereas my retail BIOS did not. Currently I'm completely stable with a custom BIOS at 1480/8GHz at max voltage.

 

[Ungari]

The cards are physically identicle, the only difference is they have a differen't(sic) BIOS flashed.

Question:

Had the review samples come with the Gaming Mode BIOS as default, do you think that any reviewer worth his salt would have published their benchmarks using this middle of the road preset, or would they have tested the cards using the OC BIOS mode to show the highest factory Core and Boost Clock available?

 

[EarthDog]

Question:

Had the review samples come with the Gaming Mode BIOS as default, do you think that any reviewer worth his salt would have published their benchmarks using this middle of the road preset, or would they have tested the cards using the OC BIOS mode to show the highest factory Core and Boost Clock available?

They would have (should have) done it as it comes out of the box.

"Worth its salt would" be to test it as it arrives out of the box. That is the fundamental problem a lot of people are having is that the review samples, mine included, come by default with the OC speeds.

View attachment 75644

My core boost clock is up to 1911MHz and MSI says boost 1797MHz
It's just the memory clock that's a bit lower on my card.

View attachment 75643

I won't try any flashing though!

Seems to me you are not in OC mode bub. That is how you get teh BASE boost of 1797. We all (should) know by now that what GPUz reports is supposed to be the MINIMUM boost, but as we all (should) know it boosts well above that in most cases.

 

[Ungari]

They would have (should have) done it as it comes out of the box.

I would have a serious problem with any reviewer that would not test the OC Mode to prove it's stability, thermals, noise, etc. for any card that features such a high boost clock.
It is after all, possibly the most salient feature these cards are offering, and to ignore that spec in the benchmarks would be an incomplete review.

 

[EarthDog]

I disagree.

Your post initially suggested they review with that preset while your follow up post now states they should check it for stability. Two completely different things.

Second, I have never EVER run across a case where the preset OC would not work on the card. Now I am sure there are some instances of it, but its not nearly enough for a reviewer to waste their time over testing it. No way is this worth my time as a reviewer to check if its stable in that mode. If this was a real concern, you would have seen people testing it and failures coming back. As it stands, nobody tests it for a reason.

Remember, boost has been implemented for generations. So even though the boost tends to go well over what is listed (you need to understand how boost works), it is still perfectly stable.

 

[Ungari]

Your post initially suggested they review with that preset while your follow up post now states they should check it for stability. Two completely different things.

Reviewing using the OC Mode preset is the same as testing it for stability as well as other factors, is it not?
What if you tested solely the Gaming Mode preset, but if it was discovered that the OC Mode preset would not maintain the high boost clocks after 15 minutes due to thermal throttling, would you not be remiss?
If I'm thinking of buying either of these cards, I want to know how it performs at the highest clock preset since there are no guarantees that I will be able to manually OC any further.

 

[EarthDog]

It is not the same to me, no. I can check for stability of a given clock without reviewing it. The reality is most reviewers don't play games for hours on end to check for stability. There are canned benchmarks they may loop a couple of times, or if there are no canned benchmarks, it's a run in the same area.

Thermal throttling will vary by user. For example, I run a bit warmer than some as I am on an open air test bench with no airFLOW. Some have a worse situation than me, some better. How can speak conclusively about thermal throttling when there are many other variables effecting that outcome? I report the temps I receive in my environment.

Do you believe no increase in voltage and a ~24 MHz clock increase are going to cause such an increase in temps?

My review will be published in a few days, but I can tell you that a 60MHz increase with a minor voltage bump (.05V) I was 1C warmer with the default fan curve. Also note, while it is a workaround, you can raise the throttling temp.

Bigger fish to fry... and if I may, I feel you are looking at the minutia but not the big picture when it comes to what you take from reviews (at least in this point).

These are my clocks without using the MSI Gaming App:
http://www.techpowerup.com/forums/t...-by-default-bioses.223571/page-3#post-3476941

You must have me on ignore, lol...

Those are gaming clocks bubba. We know what retail cards come in at.

 

[Ungari]

It is not the same to me, no.

Thermal throttling will vary by user. For example, I run a bit warmer than some as I am on an open air test bench with no airFLOW. Some have a worse situation than me, some better. How can speak conclusively about thermal throttling when there are many other variables effecting that outcome? I report the temps I receive in my environment.

Do you believe no increase in voltage and a ~24 MHz clock increase are going to cause such an increase in temps?

My review will be published in a few days, but I can tell you that a 60MHz increase with a minor voltage bump (.05V) I was 1C warmer. Also note, while it is a workaround, you can raise the throttling temp.

Bigger fish to fry... and if I may, I feel you are looking at the minutia but not the big picture when it comes to what you take from reviews (at least in this point).

I'm not trying to pick on minutia, in fact I think it is minutia to fault MSI and ASUS for putting their best foot forward by defaulting the BIOS to OC MODE for reviewers, particularly when you admitted you would use only the out of the box settings.
Mission Accomplished, ASUS and MSI! Having pointed this out, will you now review the card using the OC Mode, or will you now select the GAMING Mode BIOS for your benchmarks, or will you do both, or all three BIOS?

As to thermal testing, I think an open air bench with only the card's cooling fan(s) is the only way to be consistent as there are too many variables with chassis fans.

I think every reviewer should always test the highest advertised settings to show prospective buyers what to expect using the featured high performance specs.

 

[EarthDog]

I will review it as it came out of the box...with the OC profile. Like in my past reviews, I will continue to note the speeds I used to test it. Since MSI was kind enough to bring that BIOS out, I will also link to it and mention why it's different. Much ado about nothing for me.

 

[mouacyk]

I will review it as it came out of the box...with the OC profile. Like in my past reviews, I will continue to note the speeds I used to test it. Since MSI was kind enough to bring that BIOS out, I will also link to it and mention why it's different. Much ado about nothing for me.

I beg to differ. Your readers are not 100% all benchers. The best middle road is to use OC mode in benching software (performance) and Gaming mode in games (stability). If the middle road is not possible, I prefer you to use Gaming mode for all review testing. Crashing is detrimental to the gaming experience whereas benchers will naturally overclock anyways at said cost.

 

[EarthDog]

That is correct, they are not all benchers...in fact most are not. I dont believe I said that... and fail to see your point.

Anyway, with whatever point you are trying to make, I disagree with how they should be used. A bencher isn't going to touch that one touch overclocking software and as you said, will natually overclock anyways. So why would I bother with a 'factory' overclock for that demographic? Though I wouldn't do it anyway as everyone has a different way they use this kind of software, if I were to use it, I would use the OC Mode and Gaming mode for the gamers. Again benchers don't use this one-touch software like MSI Gaming App (they would use afterburner and push it).

 

[Ungari]

Speaking for myself, I would want to see OC Mode preset benchmarks because if I were to lose the silicon lottery I would at least know what to expect if the card was not capable of any further OC.
I recall this happening to a Newegg customer who bought an EVGA 980 Classified that was unable go beyond the factory OC.

 

[EarthDog]

Maybe an additional 1%... wow.

Im betting it would reach really high clocks on LN2, that 980 Classified... that is what it is made and binned for. People on air/water shouldn't really get those cards when one costing $30+ less reached the same clocks on air/water.

Helps to be an educated consumer and not a marketing lemming.

 

[Ungari]

Maybe an additional 1%... wow.

Im betting it would reach really high clocks on LN2, that 980 Classified... that is what it is made and binned for. People on air/water shouldn't really get those cards when one costing $30+ less reached the same clocks on air/water.

Helps to be an educated consumer and not a marketing lemming.

In the case of the Classified, some users were achieving over 1575Mhz on air, while others could not even break 1400Mhz. I do agree that the extra power phases and and such was really only useful for exotic cooling, and so I decided not to purchase one as I suffer from an irrational fear of liquid nitrogen.

 

[mouacyk]

Wizard already does manual overclocking in his reviews to show what the GPU is capable of when pushed. The part that isn't accurate is using the review sample's OC mode to test the games. There is a Gaming mode that should be proven stable along with the stated performance. If you're not proving anything you aren't educating anyone.

 

[Ungari]

Wizard already does manual overclocking in his reviews to show what the GPU is capable of when pushed. The part that isn't accurate is using the review sample's OC mode to test the games. There is a Gaming mode that should be proven stable along with the stated performance. If you're not proving anything you aren't educating anyone.

If benchmarks were shown with the middle presets juxtaposed with manual overclocking side by side I agree, but many reviewers choose not do manual overclocks as they are not indicative of a typical user's experience, and results can vary too widely. When a high factory OC is advertised, then a review of the middle preset is not useful information unless you specifically wanted to test the performance per watt which is what that preset is designed to fulfill.

 

[mouacyk]

Do you agree that the advertised Gaming preset needs to be stable for prolonged gaming? Given proper airflow if the Gaming preset is unstable, the product has failed its most important role.

The OC preset is just that. There is no stability guarantee. That is why it is misleading to test games in that mode.

 

[Ungari]

Do you agree that the advertised Gaming preset needs to be stable for prolonged gaming? Given proper airflow if the Gaming preset is unstable, the product has failed its most important role.

The OC preset is just that. There is no stability guarantee. That is why it is misleading to test games in that mode.

If an advertised OC preset is unstable in prolonged gaming I want to know this from reviewers.

 

[EarthDog]

If an advertised OC preset is unstable in prolonged gaming I want to know this from reviewers.

It never is... or if it is, its likely not something that happens often at all or if it does happen its a one off. Worry about something actually worth worrying about.

Wizard already does manual overclocking in his reviews to show what the GPU is capable of when pushed. The part that isn't accurate is using the review sample's OC mode to test the games. There is a Gaming mode that should be proven stable along with the stated performance. If you're not proving anything you aren't educating anyone.

You are caught up in marketing and not understanding what it really means it seems. Just b/c it says OC mode, doesn't mean you can't game with it. Its just a ever so slightly higher clock and fan profile adjustment.

 

[mouacyk]

If an advertised OC preset is unstable in prolonged gaming I want to know this from reviewers.

What would be the point of the Gaming preset then? OC means overclocked (as you already know) and it is unrealistic for manufacturers to guarantee prolonged stability with it.

You are caught up in marketing and not understanding what it really means it seems. Just b/c it says OC mode, doesn't mean you can't game with it. Its just a ever so slightly higher clock and fan profile adjustment.

You missed my comment about flashing the TPU msi 980 ti review sample BIOS which is unstable in 1 hour+ of game play (witcher3 and bf4). Either TPU got a cherry pick or didn't play intensive games long enough.

 

[Ungari]

What would be the point of the Gaming preset then? OC means overclocked (as you already know) and it is unrealistic for manufacturers to guarantee prolonged stability with it.



You missed my comment about flashing the TPU msi 980 ti review sample BIOS which is unstable in 1 hour+ of game play (witcher3 and bf4). Either TPU got a cherry pick or didn't play intensive games long enough.

This is very useful information concerning instability with the highest factory setting and would influence any decision to make purchases.
I recall EVGA having too many customer complaints about 980 FTW not being stable at the factory OC which was indicative of poor QC, and to EVGA's credit they allowed RMA of these cards.
If a company chooses to release a factory OC preset of their own making(which theoretically should be up to or lower than the total OC headroom which is never guaranteed) it had better stand by that presets stability or this is nothing more than a bait and switch. If reviewers chose not to deliberately test the highest boost clock advertised for stability, that would allow companies to engage in marketing shenanigans.

 

[Ungari]

What would be the point of the Gaming preset then? OC means overclocked (as you already know) and it is unrealistic for manufacturers to guarantee prolonged stability with it.

I believe the Gaming preset is a milder OC but an OC nevertheless, and that it exists for the purpose of performance to watt ratio claims, but I could be wrong.
Perhaps a better question would be; what is the purpose of the OC Mode then, if not reliable or useful?

 

[EarthDog]

With respect Mouacyk, I didn't miss it. I glossed over it and deemed it anecdotal in nature for a couple of reasons. Did it work in OC mode prior to the flash?

Some seem to believe that if a sample is tested that way, that one sample makes the rest bad so you wouldn't get it. That is pure ignorance I hate to say. Now, if 20 reviewers tested it and most borked... there would be a problem. However, that will never happen, particularly to reviewers. It is true that sometimes things are discovered that way... but testing for stability on a factory overclock option like this isn't warranted.

Its just not worth it to test like this. 99%+ of cards will overclock the 24 MHz of card we are talking about here with this card (Gaming X). Its not worth anyone's time to check like that as it really doesn't tell you much against 10K cards.

What would be the point of the Gaming preset then? OC means overclocked (as you already know) and it is unrealistic for manufacturers to guarantee prolonged stability with it.

The gaming preset is a slight overclock over stock and so is the OC... its just more aggressive. With the 1070 the difference between silent and OC is 100 Mhz while the 980Ti is 178 MHz (less % against the 980Ti).

 

[Prima.Vera]

What's the resemblance between stealing 5 cents and stealing 5 millions?

 

[mouacyk]

What's the resemblance between stealing 5 cents and stealing 5 millions?

Exactly.