MSI: *Offical* N460GTX Hawk 1GB OC Edition thread

[mikeo]

So... after posting on a number of forums, I've so far discovered.

- You can't check the voltages on the card via software (IE. GPU-z, Nvidia Inspector etc.) as it's not supported. Any values displayed by these programs are just some sort of defaults that it shows.

- Even if you could check voltages with software, you can't really compare them to voltages read direct from the card. It's like comparing apples and pears... or in this case, Apples and hovercraft!

- The only review on the Net that I can find that appears to have actually checked voltages with a multimeter, is this one:

http://www.techpowerup.com/reviews/MSI/GTX_460_HAWK/31.html

I Emailed w1zzard who wrote this article and he's been kind enough to reply. And it basically looks like my readings are within what he would expect.

And as to why the load voltage appears to jump by more than the bump voltage in Afterburner. I'll quote what someone at the MSI user forums said

"Have you taken in account that the reading point can be at the beginning of the lines.
So when you stress the card the voltage needs to be compensated by the losses from the lines.
Ergo, if more amps are needed to transport, line losses go up"

And the above is something I had wondered about and hinted at in an earlier post on this thread but did not have the right words to describe it.

So.... looks like all is OK.

Though I shall still be very interested to hear the results etc. from any other owner of the MSI HAWK edition of this card.

 

[MSI-Guru]

I am still checking into your post, and hopefully will be able to update.

I don't think you have anything to worry about in terms of the voltages though, I think it is indeed simply a case of what component(s) the voltage is being read across and the load that those components are under when reading but will see if I can get anything more.

 

[555BUK]

Question... Are they 30 min OCCT stable at those speeds?

I ran OCCT for 10mins at 930/1100 but then quit because temps went way too high (80+degrees) and I had no wish to brick my card. With the volt-modded ASUS bios the card will bench Heaven and game at 960core (1.150v) but I will not even attempt OCCT at this speed and voltage. I very much doubt that 1GHz is possible with air cooling on any GTX 460 1GB.

 

[MSI-Guru]

I ran OCCT for 10mins at 930/1100 but then quit because temps went way too high (80+degrees) and I had no wish to brick my card. With the volt-modded ASUS bios the card will bench Heaven and game at 960core (1.150v) but I will not even attempt OCCT at this speed and voltage. I very much doubt that 1GHz is possible with air cooling on any GTX 460 1GB.

Toasty ! Did you have the fan RPM maxed out? Also you open or closed case?

 

[mikeo]

I am still checking into your post, and hopefully will be able to update.

I don't think you have anything to worry about in terms of the voltages though, I think it is indeed simply a case of what component(s) the voltage is being read across and the load that those components are under when reading but will see if I can get anything more.

Much appreciated and thanks also for the "care package". Received today and proudly wearing my T-shirt (even if it is bloody freezing in sunny Scotlandshire today)

 

[mikeo]

I ran OCCT for 10mins at 930/1100 but then quit because temps went way too high (80+degrees) and I had no wish to brick my card. With the volt-modded ASUS bios the card will bench Heaven and game at 960core (1.150v) but I will not even attempt OCCT at this speed and voltage. I very much doubt that 1GHz is possible with air cooling on any GTX 460 1GB.

Fair enough. I appreciate what you say. OCCT does indeed stress a video card to an insane level (unlikely to ever hit this in normal apps / games).

Mine hit 69C / 78% fan speed (case closed) at 850MHz with 1.188V (as read at the V-sync points on the card) after 30 minute run of OCCT. So I think it does show how good / efficient the cooler is on the HAWK card.

If I can get the courage up at some point, I might see what I can get it to Kombuster stable with case open.

But happy with what I have now.
IE. Same sort of performance as a stock 470 GTX but without the heat / noise issue. Going by comparing a stock 460/470 using Anandtech's comparison utility and factoring up the figures of the stock 460 by 25% (my current OC). Though I appreciate performance scaling is not linear, but probably near enough.

Though still interested to hear any further feedback from MSI-guru and anyone else with this card.

 

[MSI-Guru]

Much appreciated and thanks also for the "care package". Received today and proudly wearing my T-shirt (even if it is bloody freezing in sunny Scotlandshire today)

Np. With your overclocking I would have thought that your area of Scotland would be a few degrees warmer

 

[mikeo]

Np. With your overclocking I would have thought that your area of Scotland would be a few degrees warmer

Think I might have to put my shorts on, it's getting so warm here now. Is that my card about to reach 4GHz? or climate change?

 

[vorticalspace]

Think I might have to put my shorts on, it's getting so warm here now. Is that my card about to reach 4GHz? or climate change?

You got MSI branded shorts as well
Is there no end to this generosity

I would have thought that for all the valuable info you have provided MSI-Guru, a nice fat discount voucher off future upgrade products should be winging your way

 

[MSI-Guru]

So... after posting on a number of forums, I've so far discovered.

- You can't check the voltages on the card via software (IE. GPU-z, Nvidia Inspector etc.) as it's not supported. Any values displayed by these programs are just some sort of defaults that it shows.

- Even if you could check voltages with software, you can't really compare them to voltages read direct from the card. It's like comparing apples and pears... or in this case, Apples and hovercraft!

- The only review on the Net that I can find that appears to have actually checked voltages with a multimeter, is this one:

http://www.techpowerup.com/reviews/MSI/GTX_460_HAWK/31.html

I Emailed w1zzard who wrote this article and he's been kind enough to reply. And it basically looks like my readings are within what he would expect.

And as to why the load voltage appears to jump by more than the bump voltage in Afterburner. I'll quote what someone at the MSI user forums said

"Have you taken in account that the reading point can be at the beginning of the lines.
So when you stress the card the voltage needs to be compensated by the losses from the lines.
Ergo, if more amps are needed to transport, line losses go up"

And the above is something I had wondered about and hinted at in an earlier post on this thread but did not have the right words to describe it.

So.... looks like all is OK.

Though I shall still be very interested to hear the results etc. from any other owner of the MSI HAWK edition of this card.

OK. In response to this, the voltage readings you see from the multimeter are correct, and the reason they are high is indeed due to the location of the V-check points. There is a certain amount of Vdrop between the V-check points and the actual GPU and memory. This means the voltage if actually measured across GPU and memory would be close to what is expected.

 

[mikeo]

OK. In response to this, the voltage readings you see from the multimeter are correct, and the reason they are high is indeed due to the location of the V-check points. There is a certain amount of Vdrop between the V-check points and the actual GPU and memory. This means the voltage if actually measured across GPU and memory would be close to what is expected.

Thanks for that. Appreciated. I'll get on and enjoy my new purchase a bit more then.

 

[mikeo]

MSI-Guru

Me again.

Out of interest. Why is it that pretty much every test / review of this card that I've read .... seems to use a different version of Afterburner than is supplied to people buying retail cards?

For example:

http://www.hardocp.com/article/2010/09/20/msi_n460gtx_hawk_edition_video_card_review/6

To quote the test article above:

"The GPU voltage slider goes from -100mV to +200mV (millivolts)"

O'no it doesn't in the version on MSI's WEB site!

Did MSI decide to reduce this to +100mV to "play safe"?

Supplying Beta software to reviewers like this doesn't look good to me. I can't help but think that this gives a false impression of this cards Over Clocking potential. Not like they got anywhere near 1GHz in the article above anyway.

Though I do accept that every GPU core has potentially different upper limits etc.
At this price point, I would have thought that MSI would have cherry picked the GPU. But I'm assuming this isn't the case.

I still think this is a decent product but that expectations have maybe been raised a little too high.

 

[MSI-Guru]

MSI-Guru

Me again.

Out of interest. Why is it that pretty much every test / review of this card that I've read .... seems to use a different version of Afterburner than is supplied to people buying retail cards?

For example:

http://www.hardocp.com/article/2010/09/20/msi_n460gtx_hawk_edition_video_card_review/6

To quote the test article above:

"The GPU voltage slider goes from -100mV to +200mV (millivolts)"

O'no it doesn't in the version on MSI's WEB site!

Did MSI decide to reduce this to +100mV to "play safe"?

Supplying Beta software to reviewers like this doesn't look good to me. I can't help but think that this gives a false impression of this cards Over Clocking potential. Not like they got anywhere near 1GHz in the article above anyway.

Though I do accept that every GPU core has potentially different upper limits etc.
At this price point, I would have thought that MSI would have cherry picked the GPU. But I'm assuming this isn't the case.

I still think this is a decent product but that expectations have maybe been raised a little too high.

Where are you downloading from? http://event.msi.com/vga/afterburner/download.htm should be the latest version.

Every new card that comes out now requires an update, and being their have been many new cards recently it means that there are a lot of different versions knocking about.

We cannot guarantee the 1GHz threshold since it is quite a difficult feat, and as you mention requires luck when it come to the GPU you have. However the platform the GPU is on tries to remove any other cause for possible limitation for overclocking besides the GPU.

Afterburner should settle down soon into some more stable versions, but remember MSI is the leader for this software and the only ones who offer this solution currently and have been trying to make it as flexible as possible in a short period of time. Stick with it

I hope you are still enjoying the card anyway

 

[HeX]

Fair enough. I appreciate what you say. OCCT does indeed stress a video card to an insane level (unlikely to ever hit this in normal apps / games).

Mine hit 69C / 78% fan speed (case closed) at 850MHz with 1.188V (as read at the V-sync points on the card) after 30 minute run of OCCT. So I think it does show how good / efficient the cooler is on the HAWK card.

You need 1.188V for 850 stable on the HAWK?

That's pretty crazy

My normal MSI does 850 @ 1.025V... =|

 

[mikeo]

You need 1.188V for 850 stable on the HAWK?

That's pretty crazy

My normal MSI does 850 @ 1.025V... =|

Please, please , please. Can people take the time to read my posts through.

You cannot compare readings via software like GPU-z, or Nvidia inspector with the readings that I am taking off the card. It's like trying to compare apples to bananas.

 

[mikeo]

Where are you downloading from? http://event.msi.com/vga/afterburner/download.htm should be the latest version.

We cannot guarantee the 1GHz threshold since it is quite a difficult feat, and as you mention requires luck when it come to the GPU you have. However the platform the GPU is on tries to remove any other cause for possible limitation for overclocking besides the GPU.

I hope you are still enjoying the card anyway

Thanks for the reply.

I picked my version up from MSI's WEB site, so it should be the same. I'm just assuming that that in the final release they dropped the max voltage increase down to +100mV. Though the README that comes with AfterBurner still says max +200mV for MSI 460 HAWK. Interesting to see that Hardcop needed +170mV to get it stable at 932MHz.

I totally accept your second comment. Though slightly disappointed that I can't get it "totally" (please note the use of this word folks) stable over 900MHz, even though the temps are fine (62C case open). Obviously my GPU is just not the best overclocker. Not too unhappy though.

And yes, I am still happy in general with my purchase. Though wish maybe that I had hung on, considering the recent price drop. The penalty you pay for being an early adopter. Though I did get a T-shirt

PS. Considering the number of folks complaining about noisy 460's in general, quite glad I bought a premium product.

 

[HeX]

Quite, but a 150mV+ discrepancy?!

 

[mikeo]

Quite, but a 150mV+ discrepancy?!

Quite.

Exactly why I've been posting like mad about it in not only this forum, but the MSI users forum, Techpowerups forum, 3D Guru (and some others that escape me).

And from discussion with the MSI rep here (who consulted his techies) and from speaking personally to W1zzard who wrote the test article about the HAWK (and many others over many, many years).

And the conclusion that I've come to, is yes "that much of a discrepancy".

To be honest... the only review that I've found where they've definitely used a multi-meter to check voltages (rather than software) was the one written by W1zzard on Techpowerup.

If you look at the review of the MSI 460 HAWK on Techpowerup's WEB site, you will indeed see that the range of voltages are in the same sort of ballpark as mine (though the site appears down for maintenance at the moment).

This does though highlight another thing that I've been trying to get at all along on this thread and on other forums. That unless you know the normal operating range of voltages (for way you are taking the readings... in this case via the V-check points on the card), then it's pretty meaningless anyway (if you see what I mean).

 

[mikeo]

You need 1.188V for 850 stable on the HAWK?

My normal MSI does 850 @ 1.025V... =|

HeX. Just out of interest. As the voltage you quote is one I've seen mentioned over and over again. And in fact GPU-z etc. even displays it for my card (IE. the HAWK) as the load voltage. I know for a fact in my case, it's just some default that the software shows.

Are you 100% sure when you increase the voltage to the card, that the figure displayed by GPU-z etc. increases in line with your voltage bump?

Also does Afterburner display a voltage readout in the monitor (if switched on obviously). Wont for the HAWK.

It's not that I'm doubting your word, just interested in confirmation from someone like yourself that obviously knows what they are doing.

 

[HeX]

I've never actually confirmed in GPU-Z or whatever, I just went by what the Afterburner setting offered me.

I know in GPU-Z my default voltage was 0.975V load.

Then went into AB and upped it until it was OCCT stable @ 850, which was 1.025V in AB.

I never actually looked in GPU-Z to confirm that this was actually being reported.


EDIT - I'm talking bobbins, 1.025V is what I needed for 800MHz, for 850MHz I needed 1.087V (max that AB allows me).


Either way, i've not checked in GPU-Z, I'll have a look see when I get in from work.