Looking for new Optiplex 790 video card

[weirdwaldo]

So I got a Dell Optiplex 790 for my birthday, Core i5 2500 3.3ghz cpu, radeon HD 6450, with a 1tb hard drive. Computer is great for everthing but gaming due to the card. Being as the computer has a 250watt PS and also would need a low profile card, are there many options available to me? I am looking for HDMI w/audio out and a fairly quiet card as it is mainly an htpc.

Any suggestions? Thanks a bunch

 

[Red Hawk]

Ouch, a 250 watt PSU really limits your choices. What resolution do you plan on playing at?

An HD 6450 is actually a decent card for quiet, power-efficient HTPC duty, but it does indeed fall short for gaming. Here are some other options:

http://www.newegg.com/Product/Produc...-380-_-Product

^^A Radeon HD 5570 with a passive cooler. $59.99 list price, $49.99 after mail-in rebate. Thanks to the passive cooler this is probably the quietest and most power-efficient, but the graphics chip itself is not very powerful. (better than the 6450 though)

http://www.newegg.com/Product/Produc...-374-_-Product

^^A Radeon HD 6570 with a passive cooler. Again, quiet and power efficient. Packs a bit more of a punch than the 5570, but it is also significantly more expensive. ($74.99)

http://www.newegg.com/Product/Produc...-871-_-Product

^^ Radeon HD 5670, actively cooled. More powerful than the 5570 and a bit better than the 6570, but also noisier and more power-hungry. Less expensive than the 6570 though -- $69.99 list price, $59.99 after mail-in rebate.

http://www.newegg.com/Product/Produc...-542-_-Product

^^Radeon HD 6670, actively cooled. This is probably the most powerful card you can get to work on your PSU. Same noise level as the 5670. $83.99 list price, $73.99 after mail-in rebate. This particular card comes with a free double-lifetime warranty -- meaning if you resell this card to someone the warranty carries over to them.

Edit: The above card has only GDDR3 memory, making it slower in some ways than the 5670 and not much better than the 6570. Here is a proper 6670 with GDDR5 memory ($96.99):

http://www.newegg.com/Product/Produc...-935-_-Product

 

[weirdwaldo]

Thanks for the swift and detailed reply my friend.

As far as resolution is concerned, im not expecting 1080p gaming at all, 720p possibly, and even then medium settings are fine by me. I do most of my gaming on the ps3/360 but there are several pc exclusives I would like to play.

Speaking of the psu being the limiter, if the card I got happened to be too muc for it, would it be bad for the card, the power supply or the motherboard?

Thanks again

 

[weirdwaldo]

Also I dont know if it matters but the 250w ps is thw Dell 90% efficient power supply, the fancy option one.

 

[Red Hawk]

You're welcome. From my understanding if the power supply can't provide the GPU enough power, the computer may randomly shut down or the screen output will fail, requiring a reboot. If it's a poor quality PSU you run the risk of frying the GPU, but I doubt your Dell PSU will have such problems even if it can't provide enough power. I have a 375 watt Dell PSU that's several years old but is quite happy powering a quad-core Intel processor and a Radeon HD 5770 (which has a significantly higher power draw than any of the cards I listed).

The "90% efficient" might make a difference. All power supplies have some degree of inefficiency, yielding lower power than they were made to. Higher quality ones have actual power yields close to their power rating.

Edit: I just noticed that the Radeon HD 6670 I listed has GDDR3 memory, not GDDR5 memory. That is bad, making it in some ways slower than the Radeon HD 5670 (which has GDDR5). Here is a proper Radeon HD 6670 with GDDR5 memory ($96.99):

http://www.newegg.com/Product/Produc...-935-_-Product

I think the best performance-for-price card is the Radeon HD 5670. The proper Radeon HD 6670 doesn't have enough performance to justify the extra $24. For a little more you can get a much faster Radeon HD 6770, after rebate, for $99.99 ($119.99 before rebate). But I'm not confident that your power supply could handle it.

 

[toyota]

that 250 watt psu makes 14 amps on the 12v I believe. there is zero chance that cheap oem psu actually makes those 14 amps under real conditions. so at best you are probably looking at 12 amps or so which is only 144 watts. considering most of the load will come from the 12v, 144 watts is not a lot to work with. plus with a cheap psu, the last thing you want to do is push it to its limits.

a system with an i5 2500 and 6670 gddr5 could easily use close to 130-140 real watts under load. my 2500 system even with an 8600gt that uses about 25-30 watts less than a 6670 gddr5 is showing 125 watts at the wall even in games that are not that demanding.

bottom line is that the card will probably work for a while but no way will that cheap psu be able to handle it after some time because that 12v line will be pushed to over 90% under full load.

 

[-Slacker-]

First make sure that you won't void your warranty for opening the case.

 

[cusideabelincoln]

my 2500 system even with an 8600gt that uses about 25-30 watts less than a 6670 gddr5

6670 and 8600GT should use about the same. There definitely is not a 25W difference.

that 250 watt psu makes 14 amps on the 12v I believe. there is zero chance that cheap oem psu actually makes those 14 amps under real conditions.

You missed this...

Also I dont know if it matters but the 250w ps is thw Dell 90% efficient power supply, the fancy option one.

my 2500 system even with an 8600gt that uses about 25-30 watts less than a 6670 gddr5 is showing 125 watts at the wall even in games that are not that demanding.

125W at the wall equates to around 105W of actual DC power, which is what power supplies are rated for delivering. The video card is also designed to run as fast as it can. In 3D games it should be drawing most of its usual power as the 2500 is certainly fast enough to push it even in slightly older, lighter games. But you're unspecific by saying "games that are not that demanding." You could be talking about Flash games or previous gen games like Splinter Cell for all we know.

 

[toyota]

6670 and 8600GT should use about the same. There definitely is not a 25W difference.

You missed this...

125W at the wall equates to around 105W of actual DC power, which is what power supplies are rated for delivering. The video card is also designed to run as fast as it can. In 3D games it should be drawing most of its usual power as the 2500 is certainly fast enough to push it even in slightly older, lighter games. But you're unspecific by saying "games that are not that demanding." You could be talking about Flash games or previous gen games like Splinter Cell for all we know.

I really doubt that is a high quality 250 watt psu but of course I could be wrong. a stock 8600gt was only a 40 watt TDP card and did not even pull but 30-35 watts so I would think its possible for there to be 25-30 watts difference and certainly even more if he got curious and oced the 6670 gddr5 a little.

and I know about what it would be at the wall. 105 REAL watts +30 for the 6670 gddr5 would be 135 REAL watts is what I was thinking. that's why I mentioned it would probably use about 130-140 watts and about 90% of whats available on the 12v based on what I mentioned earlier about likely not making but about 12 amps on it. if that psu actually makes 14 amps on the 12v in real world conditions then he should be okay. I still would not do it if it was my pc though.

as for games not that demanding, of course you know it was not something like Crysis for me to say that. it was mainly RE 5 on medium settings at 1280x960 where I kept looking at the wattage. I did check some other games after my comment and 120-125 at the wall seems to be the most it will pull. most games were only 105-110 watts most of the time but would have a few spikes above that.

 

[cusideabelincoln]

I really doubt that is a high quality 250 watt psu but of course I could be wrong. a stock 8600gt was only a 40 watt TDP card and did not even pull but 30-35 watts so I would think its possible for there to be 25-30 watts difference and certainly even more if he got curious and oced the 6670 gddr5 a little.

and I know about what it would be at the wall. 105 REAL watts +30 for the 6670 gddr5 would be 135 REAL watts is what I was thinking.

I think your estimates are a bit off for the 8600GT and 6670. They are going to use about the same; fluctuations will might put a 10W difference between them but definitely not 30W. The 8600GT should use close to its rated TDP. The 6670's 66W TDP is an overestimation.

6670 using less than the 5670
5670 using a bit more than the GT 220 - should put the 6670 and 220 on equal ground
GT 220 using a bit more than the 9500GT
9500GT using less than the 8600GT - should put the 220 and 8600GT on equal ground, and consequently the 6670 too.

Forgive multiple reviews, but there is no review containing them both since the 8600GT is so old. TPU uses consistent tests and copy+pastes most of his data between reviews, so I think this is the fairest estimate you're going to find.

 

[toyota]

I think your estimates are a bit off for the 8600GT and 6670. They are going to use about the same; fluctuations will might put a 10W difference between them but definitely not 30W. The 8600GT should use close to its rated TDP. The 6670's 66W TDP is an overestimation.

6670 using less than the 5670
5670 using a bit more than the GT 220 - should put the 6670 and 220 on equal ground
GT 220 using a bit more than the 9500GT
9500GT using less than the 8600GT - should put the 220 and 8600GT on equal ground, and consequently the 6670 too.

Forgive multiple reviews, but there is no review containing them both since the 8600GT is so old. TPU uses consistent tests and copy+pastes most of his data between reviews, so I think this is the fairest estimate you're going to find.

yeah different sites will have different results of course based on how they are testing so it is hard to make a direct comparison. even in that techpowerup link of yours it shows the 6670 hitting a peak of 61 watts. xbit labs had the 8600gts which of course uses more power than an 8600gt using only 47 watts max. so again there certainly could be 25-30 watts differences between a stock 8600gt and 6670 gddr5.

 

[Red Hawk]

Wait, why would a 6670 use less power than a 5670? It uses the same process, and has more shaders than the 5670.

 

[cusideabelincoln]

even in that techpowerup link of yours it shows the 6670 hitting a peak of 61 watts. xbit labs had the 8600gts which of course uses more power than an 8600gt using only 41 watts max.

Can't compare across different sites. I've never completely trusted Xbit's older numbers either, as they are drastically different. Both TPU and Xbit have tested GPU load of just the +12V, and I have seen radically different numbers for the same cards. So you can't compare them. Here's just one example

HD 4770 review - Compare to the 4770 measurements in the TPU reviews. Xbit measures a peak of 49W while TPU measures a peak of 69W. I bet if Xbit used that same methodology in testing the 6670 as the 8600GT the numbers would come out very, very close. But comparing across different sites is not a good idea. Meanwhile while TPU has multiple reviews, he has been using the same exact methodology for 3 out of the 4 reviews I linked. Only in the newest did he switch, but his rankings for cards are still consistent with the old method.

so again there most certainly could be 25-30 watts differences between a stock 8600gt and 6670 gddr5.

Nope. You can't subtract 61W from 41W because those numbers are from different websites using different equipment in different tests. Your conclusion is inconclusive at best when trying to do that. And I've also just demonstrated with the 4770 results that Xbit usually measures less than TPU either by a little or by a lot.

Wait, why would a 6670 use less power than a 5670? It uses the same process, and has more shaders than the 5670

Mature process and the fan power consumption can have a significant (&#37 impact on these low power cards. That's probably the difference.

 

[toyota]

it can easily be in the 50-60 watt range for the 6670 gddr 5 and be in the 30-35 watt range for the 8600gt. so again it "could" be 25-30 watts difference. you cannot prove that there will not be and I can not prove that there will. its always best to assume realistic worst possible cases when making sure a psu is up to the job.

 

[cusideabelincoln]

I just [redacted] proved they use the same and I disproved your entire [redacted] logic of trying to compare numbers across different reviews.

Want even more proof of the difference between review sites?

Here's xbit's 5670 review

They measure the 5670's peak 3D game laod at 28W. This is a card with a 61W TDP. That's less than the numbers you've thrown around they themselves measured for the 8600GT. That's also less than the numbers TPU measured for the 5670.

It's easier to make an argument when you're not swearing. Please keep the profanity out of our technical forums
-ViRGE

 

[toyota]

I just [redacted] proved they use the same and I disproved your entire [redacted] logic of trying to compare numbers across different reviews.

sounds like you have a temper problem...

if you want to think there is no chance there could ever be 25-30 watts difference then fine. I still think it could be possible in some cases. I could buy a 6670 and see but if it did turn out to be close to that, you would only blame my testing.

 

[cusideabelincoln]

You have a problem.

Want even more proof of the difference between review sites?

Here's xbit's 5670 review

They measure the 5670's peak 3D game laod at 28W. This is a card with a 61W TDP. That's less than the numbers you've thrown around they themselves measured for the 8600GT. That's also less than the numbers TPU measured for the 5670.

 

[cusideabelincoln]

You can't compare exact data across multiple review sites. The only thing you can compare is relative performance. How hard is that to understand? Yet you keep trying to do it.

 

[toyota]

You can't compare exact data across multiple review sites. The only thing you can compare is relative performance. How hard is that to understand? Yet you keep trying to do it.

I understand that fully. and you cant seem to accept that the tests done at different times even on the same sites, and even cards themselves can vary a few watts. its NOT impossible to end up in a scenario where 25-30 watts could be the difference.

 

[cusideabelincoln]

I understand that fully. and you cant seem to accept that the tests done at different times and even cards themselves can vary a few watts. its NOT impossible to end up in a scenario where 25-30 watts could be the difference.

Except there is not a 25-30W difference. You have absolutely no evidence at all to support this.

TPU only takes measurements once, and then copies those measurements from review to review. You can compare relative performance in this manner using their reviews.

You cannot compare the two sites directly without a point of reference.

Just accept you're wrong, because you flat out are. I doubt you will because of your huge ego.


Re: "I doubt you will because of your huge ego"
Let's not include personal attacks here, especially within a heated discussion.

Moderator jvroig

 

[toyota]

Except there is not a 25-30W difference. You have absolutely no evidence at all to support this.

TPU only takes measurements once, and then copies those measurements from review to review. You can compare relative performance in this manner using their reviews.

You cannot compare the two sites directly without a point of reference.

Just accept you're wrong, because you flat out are. I doubt you will because of your huge ego.

how about this? and I hate to burst your bubble but I used a 4670 and 8600gt in my old 5000 X2 pc. the stock 4670 used over 20 watts more than the 8600gt which was slightly oced at the time. go and look at Anandtech 5670 review and you will see that a 5670 uses 21 more watts than 4670. now we know a 6670 only uses just a little bit less than a 5670. its a fact that the 6670 uses more power than a 4670 according to Anandtech. that right there shows a 6670 would be using around 25 watts more than my 8600gt when it was oced a little. now whats your excuse?


http://www.anandtech.com/show/2917/13

 

[cusideabelincoln]

Anand is using a synthetic Furmark for those tests, which is not indicative of 3D gaming performance at all. And now you're just comparing your hard data to Anandtech's hard data? What a [redacted] joke.

Load power paints a different picture however, and this is where the concept of a “power virus” pokes its head up again.

The only thing I'll concede is there could be such a difference you claim when using synthetic tests. I do not concede it happens in gaming situations.


No profanity in the tech forums, please.

Moderator jvroig

 

[toyota]

Anand is using a synthetic Furmark for those tests, which is not indicative of 3D gaming performance at all. And now you're just comparing your hard data to Anandtech's hard data? What a [redacted] joke.

see how you have an excuse for everything?


according to anandtech and me...
FACT: a 4670 uses about 20 more watts than an 8600gt
FACT: a 6670 uses more power than a 4670

so again it was possible to use 25 watts or more over the 8600gt just like I said. you ALWAYS have to jump on me in nearly every thread and nit pick my comments. in the end I was trying to give him the best advice and it turns out I was pretty much right.

 

[cusideabelincoln]

see how you have an excuse for everything?

FACT: a 4670 uses about 20 more watts than an 8600gt
FACT: a 6670 uses more power than a 4670

so again it was possible to use 25 watts or more over the 8600gt just like I said.

It's not a [redacted] excuse; it's the real world.

Load power paints a different picture however, and this is where the concept of a “power virus” pokes its head up again.

The only thing I'll concede is there could be such a difference you claim when using synthetic tests. I do not concede it happens in gaming situations.

We're talking about 3D games here, not Furmark and not synthetic loads. It's only possible under synthetic loads. WHICH IS [redacted] POINTLESS TO THE OP. BUT YOU WIN THE ARGUMENT!


No profanity in the tech forums, please.

Moderator jvroig

 

[toyota]

and I was talking about REAL games right here in front of me. there was literally around 20 watts difference between the 4670 and 8600gt. if my 8600gt would have been stock at the time then it would have been probably 25 watts difference in games. and AGAIN the 6670 clearly uses more power than a 4670 even in real games...