GTX 560ti Users: Information Needed


log in

Advanced search

Message boards : Number crunching : GTX 560ti Users: Information Needed

Previous · 1 · 2 · 3 · 4 · 5 · Next
Author Message
Profile HAL9000
Volunteer tester
Avatar
Send message
Joined: 11 Sep 99
Posts: 4443
Credit: 119,092,632
RAC: 140,447
United States
Message 1128065 - Posted: 15 Jul 2011, 22:24:29 UTC - in response to Message 1128058.

My new PSU has ferrite chokes on the PCIe 12V connectors at the GPU end. I take that as a big warning sign that GPUs might be capable of trashing the 12V rails beyond load regulation limits of some power supplies

Or it's just more advertising gimmickry.
Chokes down there will just stop electrical noise (of the appropriate frequency) from the video card making it's way back into the PSU. Or any noise picked up in the long power leads from getting into the video card.
Or most likely, it looks impressive.

In any case, it can't hurt....and Corsair must have had a reason for adding them other than looking cool. Ferrite is pretty cheap. Might have to do some Googleing on this to see if there is any other info out there.

Or they started doing it for marketing reasons. We now include them with all of our products. Some customers complained that we were only including them with the larger devices. Which was due to a silly reason they were only needed there.

I could see them being needed for some of those high end cards tho. With the massive DC to DC supplies the high end cards have on there they probably start kicking out a bit of noise. I would hope the card manufacture would take care of that on the card.
____________
SETI@home classic workunits: 93,865 CPU time: 863,447 hours

Join the BP6/VP6 User Group today!

Profile HAL9000
Volunteer tester
Avatar
Send message
Joined: 11 Sep 99
Posts: 4443
Credit: 119,092,632
RAC: 140,447
United States
Message 1128071 - Posted: 15 Jul 2011, 22:50:19 UTC - in response to Message 1128069.


I could see them being needed for some of those high end cards tho. With the massive DC to DC supplies the high end cards have on there they probably start kicking out a bit of noise. I would hope the card manufacture would take care of that on the card.

You would hope.......
Just think. By adding a few cents worth of ferrite to the boards, they could advertise......
"Our new DC input stabilization circuitry insures better computer stability at either stock or highly overclocked speeds."

But space is really at a premium on most high end cards, and one thing that ferrite is NOT, is small. It takes a certain amount of the material to have enough effect.

Although I could see an aftermarket for plug in ferrite filter adapters that would go between the card and the PSU connecting plug.

Looking at my cards I do see a fair number of those 1/2" square ferrite chokes on the PCB. So it looks like Sapphire is taking some steps. How effective they are remains to be seen. I might roll out my old HP scope and have a look.
____________
SETI@home classic workunits: 93,865 CPU time: 863,447 hours

Join the BP6/VP6 User Group today!

Profile HAL9000
Volunteer tester
Avatar
Send message
Joined: 11 Sep 99
Posts: 4443
Credit: 119,092,632
RAC: 140,447
United States
Message 1128140 - Posted: 16 Jul 2011, 1:59:43 UTC - in response to Message 1128079.

Could be interesting....but be careful, pretty crowded on those boards, easy to short something out.

So watch where you're stickin' yer probe.

Hmmm yeah there is a bit of noise getting generated by the GPUs.

HP 120B scope at 2ms & 100mv. ATI HD4870 & Enermax 525w Pro82+. Readings taken off of a spare PCIe connector. Click on them for larger versions.

____________
SETI@home classic workunits: 93,865 CPU time: 863,447 hours

Join the BP6/VP6 User Group today!

Profile HAL9000
Volunteer tester
Avatar
Send message
Joined: 11 Sep 99
Posts: 4443
Credit: 119,092,632
RAC: 140,447
United States
Message 1128158 - Posted: 16 Jul 2011, 2:54:09 UTC - in response to Message 1128150.

Could be interesting....but be careful, pretty crowded on those boards, easy to short something out.

So watch where you're stickin' yer probe.

Hmmm yeah there is a bit of noise getting generated by the GPUs.

HP 120B scope at 2ms & 100mv. ATI HD4870 & Enermax 525w Pro82+. Readings taken off of a spare PCIe connector. Click on them for larger versions.

So this is already coming out of the GPU and going through the PSU to a spare output connector?
Would be interesting to see what adding some ferrite beads to the PCI-E power connectors at the GPU would do.....

I did some looking around, but I cannot find any supplier so far that has picked up on this phenomenon and is producing a plug-in adapter for this.
If I had the time and money, I would jump on this right away. You could produce these things for real cheap....and what geek in his own right mind would not spend $5.00 each for a couple to plug into their prized GPUs?
You'd have to do some research, of course, to get the right ferrite that worked at the frequencies involved.

Amidon is a company I dealt with years ago that offered ferrite beads that responded at different frequencies.
It appears they are still in business and offer online ordering if anybody wants to play 'engineer for a day'.

All of my PCIe power cables are the 6+2P connector type. The card has two 6P connectors. So I connected to the spare +2 bit. All 4 power wires on each connector run back to the supply. Touching the probe to the solder joints on the back of the card gave very similar readings, but lacked anything to clip onto so I could take a photo.
Also I did make sure this wasn't just PSU noise by loading up the CPU & GPU separately.
____________
SETI@home classic workunits: 93,865 CPU time: 863,447 hours

Join the BP6/VP6 User Group today!

Profile jason_gee
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 24 Nov 06
Posts: 5051
Credit: 73,852,693
RAC: 12,146
Australia
Message 1128159 - Posted: 16 Jul 2011, 3:03:00 UTC - in response to Message 1128140.

HP 120B scope at 2ms & 100mv. ATI HD4870 & Enermax 525w Pro82+. Readings taken off of a spare PCIe connector. Click on them for larger versions.


Hmm Interesting, thanks. No idea what effect those high frequency spikes would do, I'd imagine not much directly, but it certainly explains why the ferrites might be there on the new Ax1200. Aside from the peaks, the ripply harmonic looking bits look like about half a division period, so ~1ms, or 1KHz. Capacitors would age faster than if those ripply bits weren't there I reckon. Might explain some of the reasons why higher grade & capacity PSUs are recommended though the GPUs themselves don't draw that much.

Yeah probably we're headed into geek territory alright, where it'll matter only to those trying to squeeze every bit of performance out, but it's good to know there are reasons for the PSU differences that can be measured.

Jason

____________
"It is not the strongest of the species that survives, nor the most intelligent that survives. It is the one that is the most adaptable to change."
Charles Darwin

Profile jason_gee
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 24 Nov 06
Posts: 5051
Credit: 73,852,693
RAC: 12,146
Australia
Message 1128160 - Posted: 16 Jul 2011, 3:07:36 UTC - in response to Message 1128158.
Last modified: 16 Jul 2011, 3:08:28 UTC

...Also I did make sure this wasn't just PSU noise by loading up the CPU & GPU separately.


No worries there, the scope shots make it pretty clear the GPU does contaminate the rails, which is what I was after querying the presence of ferrites on my PCIe connectors. I'd have to look up at some stage to see to what degree that amount of ripple on the 12V rail pushes the ATX spec... Seems like a lot to me remembering Seasonic and Corsair reviews indicating ~10mV ripple under load, but lifespan & stability effects aside I think it is within spec. I'd imagine with newer bigger cards the effect could be more drastic though.
____________
"It is not the strongest of the species that survives, nor the most intelligent that survives. It is the one that is the most adaptable to change."
Charles Darwin

Profile HAL9000
Volunteer tester
Avatar
Send message
Joined: 11 Sep 99
Posts: 4443
Credit: 119,092,632
RAC: 140,447
United States
Message 1128162 - Posted: 16 Jul 2011, 3:18:11 UTC - in response to Message 1128159.

HP 120B scope at 2ms & 100mv. ATI HD4870 & Enermax 525w Pro82+. Readings taken off of a spare PCIe connector. Click on them for larger versions.


Hmm Interesting, thanks. No idea what effect those high frequency spikes would do, I'd imagine not much directly, but it certainly explains why the ferrites might be there on the new Ax1200. Aside from the peaks, the ripply harmonic looking bits look like about half a division period, so ~1ms, or 1KHz. Capacitors would age faster than if those ripply bits weren't there I reckon. Might explain some of the reasons why higher grade & capacity PSUs are recommended though the GPUs themselves don't draw that much.

Yeah probably we're headed into geek territory alright, where it'll matter only to those trying to squeeze every bit of performance out, but it's good to know there are reasons for the PSU differences that can be measured.

Jason

This is my most power hungry card at 150w. I am guessing with cards that draw more power it might be even more pronounced.

I couldn't lock onto the ripples with the triggering on my scope. The best I was able to do was getting them to run across the screen slow enough for a photo. Overall it doesn't look to bad for me, but it does look like there is a funky harmonic generating 300mv spikes around 150-160Hz. I might have to do something about it now that I know it is there...
____________
SETI@home classic workunits: 93,865 CPU time: 863,447 hours

Join the BP6/VP6 User Group today!

Profile jason_gee
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 24 Nov 06
Posts: 5051
Credit: 73,852,693
RAC: 12,146
Australia
Message 1128164 - Posted: 16 Jul 2011, 3:31:38 UTC - in response to Message 1128162.
Last modified: 16 Jul 2011, 3:40:33 UTC

... Overall it doesn't look to bad for me, but it does look like there is a funky harmonic generating 300mv spikes around 150-160Hz. I might have to do something about it now that I know it is there...


LoL, I do think you're well within ATX (baseline) specs, though reducing that wouldn't hurt longevity/stability I believe.

From some ATX specs in Intel PSU design guides (considered as minimum from our point of view I guess):

12V1 rail: 12V +/- 5% at all times , So a variation of +/- 600mV is allowed
But: Outputs must remain within the regulation limits with load changing repetition rates of 50Hz -> 10KHz, with simultaneous load changes on all rails in the same direction. I wager more than a few 'average' PSUs would let at least those spikes go beyond spec with a heavy, rapidly changing load like that.

[Edit:] Slew rates (1 Amp/usec) & other transient characteristics seem to be specified too, but I've never actually seen PSU specs list that data...

Jason
____________
"It is not the strongest of the species that survives, nor the most intelligent that survives. It is the one that is the most adaptable to change."
Charles Darwin

Profile HAL9000
Volunteer tester
Avatar
Send message
Joined: 11 Sep 99
Posts: 4443
Credit: 119,092,632
RAC: 140,447
United States
Message 1128165 - Posted: 16 Jul 2011, 3:39:19 UTC - in response to Message 1128163.



I couldn't lock onto the ripples with the triggering on my scope. The best I was able to do was getting them to run across the screen slow enough for a photo. Overall it doesn't look to bad for me, but it does look like there is a funky harmonic generating 300mv spikes around 150-160Hz. I might have to do something about it now that I know it is there...

LOL....
NOW you are talking geeky....and it's gonna bug the heck out of you now that you know it's there.

Then I did something stupid. I started looking at the outputs on my other machines. *shivers* I've put my scope away now "there is no ripple...there is no ripple..."
____________
SETI@home classic workunits: 93,865 CPU time: 863,447 hours

Join the BP6/VP6 User Group today!

Profile jason_gee
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 24 Nov 06
Posts: 5051
Credit: 73,852,693
RAC: 12,146
Australia
Message 1128168 - Posted: 16 Jul 2011, 3:46:26 UTC - in response to Message 1128167.
Last modified: 16 Jul 2011, 3:50:53 UTC

And geezus, Slavac.
Sorry if we took your thread so far off course.


He started it :P

http://www.formfactors.org/developer/specs/ATX12V%20PSDG2.01.pdf

I'll be doing some light reading. Probably will have to start making Lunatics brand PSUs now, LoL

[Edit:] Uh Oh
Output
Max. Ripple & Noise
(mVpp)
+12 V1DC 120
+12 V2DC 120
+5 VDC 50
+3.3 VDC 50
-12 VDC 120
+5 VSB 50

____________
"It is not the strongest of the species that survives, nor the most intelligent that survives. It is the one that is the most adaptable to change."
Charles Darwin

Profile jason_gee
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 24 Nov 06
Posts: 5051
Credit: 73,852,693
RAC: 12,146
Australia
Message 1128170 - Posted: 16 Jul 2011, 3:55:49 UTC - in response to Message 1128169.

...(240v 100 amp circuit required)"


LoL, Nuclear reactor not included.

____________
"It is not the strongest of the species that survives, nor the most intelligent that survives. It is the one that is the most adaptable to change."
Charles Darwin

Profile jason_gee
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 24 Nov 06
Posts: 5051
Credit: 73,852,693
RAC: 12,146
Australia
Message 1128175 - Posted: 16 Jul 2011, 4:02:22 UTC - in response to Message 1128173.

...(240v 100 amp circuit required)"


LoL, Nuclear reactor not included.

And I don't think any solar cells we know of now are gonna cut it for this baby.


Not only does it come with a velvet bag & with handy modular cable pouch, but also a tin foil hat, & a morse code telegraph printer so you can detect all those triplets the stock app chokes on at the moment!
____________
"It is not the strongest of the species that survives, nor the most intelligent that survives. It is the one that is the most adaptable to change."
Charles Darwin

Previous · 1 · 2 · 3 · 4 · 5 · Next

Message boards : Number crunching : GTX 560ti Users: Information Needed

Copyright © 2014 University of California