ATI Radeon HD 5450 & HD 5570 Graphics Cards
-
- Posts: 103
- Joined: Fri Jan 08, 2010 10:30 pm
nice review mike. The 5450 sounds like a perfect replacement for the aging x1250 intergreted graphics in my htpc.
Is it possible that the 5450's low power draw could be less then the internal graphics on my Asus M2A-VM? Or would there be no offset to the internal graphics being disabled?
My HTPC uses a 90 Watt PicoPSU with an athlon x2 2.0ghz at 1.05 volts. (65nm). And laptop hardrive with 2 gigs of ram. So i dont think that is too much for a Pico, but still dont care to go over budget.
Is it possible that the 5450's low power draw could be less then the internal graphics on my Asus M2A-VM? Or would there be no offset to the internal graphics being disabled?
My HTPC uses a 90 Watt PicoPSU with an athlon x2 2.0ghz at 1.05 volts. (65nm). And laptop hardrive with 2 gigs of ram. So i dont think that is too much for a Pico, but still dont care to go over budget.
-
- Posts: 310
- Joined: Mon Sep 03, 2007 1:45 pm
- Location: Edmonton, Alberta, Canada
Wow, SPCR got a 55x0 series card before Anand or Tom's. Looks like SPCR is really moving up in the world. Well deserved, Mike! Keep up the good work!
EDIT: Haha, looks like it was just the NDA keeping the others from publishing....
EDIT: Haha, looks like it was just the NDA keeping the others from publishing....
Last edited by hybrid2d4x4 on Tue Feb 09, 2010 10:04 am, edited 1 time in total.
I highly doubt 7w max load is going to put you over capacity on your Pico-90, as long as your power brick is rated 90+ watts. You might invest in a Kill-a-watt or similar device to make sure.vortex222 wrote:My HTPC uses a 90 Watt PicoPSU with an athlon x2 2.0ghz at 1.05 volts. (65nm). And laptop hardrive with 2 gigs of ram. So i dont think that is too much for a Pico, but still dont care to go over budget.
-
- SPCR Reviewer
- Posts: 1115
- Joined: Fri Mar 04, 2005 9:07 pm
- Location: Vancouver
-
- Posts: 316
- Joined: Thu Aug 10, 2006 11:07 am
Since SPCR linked to Anandtech's review of the 5450, I am assuming that they are aware of the 5450's shortcomings in deinterlacing video. As ATI (unlike nvidia) uses its stream processors to decode video, the 80-shader 5450 can't keep up with the full suite of options in the Catalyst Control Center.
This should be brought up in any discussion of the 5450 as a HTPC card (in fact, its only real niche).
Yet there is no mention of it here?
This should be brought up in any discussion of the 5450 as a HTPC card (in fact, its only real niche).
Yet there is no mention of it here?
For decoding video on the graphics card (DXVA), I think it is either on or off.Voldenuit wrote:Since SPCR linked to Anandtech's review of the 5450, I am assuming that they are aware of the 5450's shortcomings in deinterlacing video. As ATI (unlike nvidia) uses its stream processors to decode video, the 80-shader 5450 can't keep up with the full suite of options in the Catalyst Control Center.
For deinterlacing, I think the 5450 might have enough shaders, but that deinterlacing test video that anand used is just over the top and unrealisticly difficult. It may work fine on normal video with vector adaptive activated all the time. It just switches to a lesser deinterlacer when the going gets tough. That's my understanding of it anyways.
-
- Posts: 2198
- Joined: Thu Feb 10, 2005 11:20 am
- Location: TN, USA
There should be a disclaimer about the 4670 idle power draw. Numerous other sites got higher idle numbers than SPCR on that card.
For convenience instead of linking to a bunch of reviews I'll just link to TTAGPR.
For those not wanting to decipher insanely large charts I'll distill a few numbers into text:
Idle power 3D
5770 18W 78W
5750 13W 54W
5670 13W 51W
5570 10W 43W
5450 7W 15W
4770 26W 58W
4750 29W 55W
4670 9W 48W
4650 9W 30W
4550 7W 18W
4350 4W 15W
SPCR shows the 4670 vs 5570 idle comparison as 3W vs 8W but TTAGPR shows that comparison as 9W vs 10W.
Blame that on measurement error, sample variance (manufacturing variance), non reference vs reference cards, or any other factor you like but someone just buying a random 4670 video card is unlikely to see the idle power draw you are quoting here.
For convenience instead of linking to a bunch of reviews I'll just link to TTAGPR.
For those not wanting to decipher insanely large charts I'll distill a few numbers into text:
Idle power 3D
5770 18W 78W
5750 13W 54W
5670 13W 51W
5570 10W 43W
5450 7W 15W
4770 26W 58W
4750 29W 55W
4670 9W 48W
4650 9W 30W
4550 7W 18W
4350 4W 15W
SPCR shows the 4670 vs 5570 idle comparison as 3W vs 8W but TTAGPR shows that comparison as 9W vs 10W.
Blame that on measurement error, sample variance (manufacturing variance), non reference vs reference cards, or any other factor you like but someone just buying a random 4670 video card is unlikely to see the idle power draw you are quoting here.
Last edited by dhanson865 on Mon Feb 08, 2010 7:26 am, edited 1 time in total.
-
- Posts: 2198
- Joined: Thu Feb 10, 2005 11:20 am
- Location: TN, USA
Thanks for posting those power consumption numbers. They seem to be more accurate -- or at least they jibe better with what I've seen on other sites. The 1 W difference under idle versus load for the HD 5450 just didn't seem right at all. SPCR might need to look into their testing methods a little more closely.
-
- Posts: 273
- Joined: Tue Aug 04, 2009 7:03 am
- Location: Brooklyn, NY
Still Not the Perfect HTPC Card according to Anand Tech: http://www.anandtech.com/video/showdoc.aspx?i=3738&p=3
Little NDA date confusion at SPCR?
Little NDA date confusion at SPCR?
-
- Posts: 404
- Joined: Sun Jul 04, 2004 9:18 pm
- Location: Illinois, USA
-
- Posts: 2198
- Joined: Thu Feb 10, 2005 11:20 am
- Location: TN, USA
Actually I trust SPCRs testing methods more than other sites. If they make a mistake they always own up to it and learn from it improving their reviews year after year.blaster5k wrote:Thanks for posting those power consumption numbers. They seem to be more accurate -- or at least they jibe better with what I've seen on other sites. The 1 W difference under idle versus load for the HD 5450 just didn't seem right at all. SPCR might need to look into their testing methods a little more closely.
Something was different about the 4670 they tested vs the 4670 most of us could buy. That isn't an issue with how they test just an issue of what they tested.
As to the 5450 I wonder if they ran into the ATI furmark coverup again. See http://www.geeks3d.com/20080826/amdati- ... ark-proof/ for just one past example. (for the record Nvidia has done this in their drivers as well).
It would be interesting to know if ATI Catalyst 8.69 RC3 limits power draw on a 5450 with furmark.exe but I'm ok either way the power draw is so low even on other 3D loads it doesn't matter to me if it is 7W or 15W in 3D apps.
I generally trust SPCR's power measurements more than other sites too. I'm not sure some of the gaming-oriented ones even know the difference between AC and DC.
In this case though, something is definitely up. Most sites don't run CPUBurn at the same time as FurMark, so I wonder if that has something to do with it. Of course, how much these things get throttled under FurMark is an unknown too.
Anyway, the difference between the rated 19.1 W load and 7 W load that SPCR got is potentially big for those us using smaller power supplies.
In this case though, something is definitely up. Most sites don't run CPUBurn at the same time as FurMark, so I wonder if that has something to do with it. Of course, how much these things get throttled under FurMark is an unknown too.
Anyway, the difference between the rated 19.1 W load and 7 W load that SPCR got is potentially big for those us using smaller power supplies.
-
- Posts: 273
- Joined: Tue Aug 04, 2009 7:03 am
- Location: Brooklyn, NY
ht4u.net does actual direct measurements of video card power usage (using a PCIe extender card). Here's typical charts of idle and load power usage, including the 5450 & 5670: http://ht4u.net/reviews/2010/amd_radeon ... ndex13.php
You'll be able to figure it out without knowing German.
You'll be able to figure it out without knowing German.
Wow, that ht4u link was nice. According to their numbers, I could get a whopping 25W idle reduction in DC, if I replaced my Nvidia 8800GTS 512MB by an ATI 5870 and while my system uses around 110W AC at idle now and my Corsair VX450 being about 80% efficient at that load, would reduce my idle by 31W AC to under 80W.
This would already make my 1m^2 walk-in closet containing the computer much cooler. At the moment it gets a bit warmer in there than would be good for the system in the long term.
And of course I would not get anything just for reduction in power, but I can't play GTA 4 on my system without making it look like lego blocks.
I guess I'm still waiting for the new Nvidia releases just to get a 5870 cheaper.
This would already make my 1m^2 walk-in closet containing the computer much cooler. At the moment it gets a bit warmer in there than would be good for the system in the long term.
And of course I would not get anything just for reduction in power, but I can't play GTA 4 on my system without making it look like lego blocks.
I guess I'm still waiting for the new Nvidia releases just to get a 5870 cheaper.
at the wall my UPS stated its consuming ~55 watts idle and 75 watts when spinning up a DVD and playing a video. So actual DC consumption should be much lower. I just have not cared to run a Pico at its peak capability. Also i take those numbers with a grain of salt.Meato wrote:I highly doubt 7w max load is going to put you over capacity on your Pico-90, as long as your power brick is rated 90+ watts. You might invest in a Kill-a-watt or similar device to make sure.vortex222 wrote:My HTPC uses a 90 Watt PicoPSU with an athlon x2 2.0ghz at 1.05 volts. (65nm). And laptop hardrive with 2 gigs of ram. So i dont think that is too much for a Pico, but still dont care to go over budget.
Anandtech has a follow-up blog that notes the 5570 with an updated driver has enough power for all of the 5xxx series video processing. The 5450 has enough power for deinterlacing at 1920x1080 (although not enough for other post-processing effects).
http://anandtech.com/weblog/showpost.aspx?i=669
http://anandtech.com/weblog/showpost.aspx?i=669