Video Card UpGrade

[Masherbrum]

Lusche, if you are mostly using the new video card for AH, I think the 9600GT will match better with your CPU since it will be the limiting factor. On less CPU intense games you may get more fps benefits from a 9800GT.

Your PSU will handle a 9800GT fine.

With a Minimum of 400 watts and 26 amps required for most 9800's, I'd strongly disagree.  On paper that gives him 3 amps and 50W headroom, that card would be toast within a year.

[humble]

The problem is identifying the true draw. I've seen review/tests with as low as 230W (roughly 19@) and figures as high as yours. The problem is that at 70% efficiency your only getting 315W of guaranteed continuous clean power (assuming the PS actually meets the label). If you look at the label it specifies 145W for the lower voltage rails....leaving 305 for the 12V rails. That gives a theoretical max of just under 26 amps....but at our 70% number we're under 19@ total...

[BaldEagl]

29 amps on the combined 12V rails is marginal for a 9800.  I wouldn't try it. 

Lusche, get the 9600.  You'll be able to run full bore in the game and save the rest for a new system down the road.

[Lusche]

Thanks for all the input.

Yes, I will stick with the  9600. The 9800 is simply pushing it too far in regards to power consumption while not being that much better considering my CPU and the resolution im playing at. Also I'm playing mostly AHII, I originally planned to get Rise of Flight, but their DRM and upgrade policy made this game a no-go for me anyway, no matter what system I have.

[Boozeman]

Well, do you guys actually know what a 9800GT consumes in RL, not on "paper"?

As a worst case scenario, which would be running Furmark, a 8800GT (the 9800GT is essentially the same card) draws about 112 W.

> http://ht4u.net/reviews/2009/sapphire_radeon_hd_4770/index10.php 

Now, Lusches PSU can deliver a total of 350 W on the 12V rails... so, roughly 300% more. That is a substantial overhead. So much in fact that I doubt the PSU will even run in the load range where it is most efficient.

But I agree, the 9600GT fits his CPU better, unless if he wants to maximise AA and AF settings.

BTW, did you know that the 9600GT consumes on average only 10% less power then the 9800GT? So if a 9800GT is a no-go, the 9600GT is too.   

 

[BaldEagl]

From eVGA's web-site:

9800 GTX+
Requirements
Minimum of a 450 Watt power supply.
(Minimum recommended power supply with +12 Volt current rating of 24 Amps.)

9800 GT
Requirements
Minimum of a 400 Watt power supply.
(Minimum recommended power supply with +12 Volt current rating of 26 Amps.)

9600 GT Low Power
Requirements
Minimum of a 300 Watt power supply.
(Minimum recommended power supply with +12 Volt current rating of 18 Amps.)


That looks like a lot more than 10% to me.

I know you don't believe in the manufacturer's recommendations as has been evidenced by your posts in the past but I'm pretty sure that they don't just make this stuff up.  There has to be adequate overhead because a video card isn't the only thing running on a machine.

[humble]

Well, do you guys actually know what a 9800GT consumes in RL, not on "paper"?

As a worst case scenario, which would be running Furmark, a 8800GT (the 9800GT is essentially the same card) draws about 112 W.

> http://ht4u.net/reviews/2009/sapphire_radeon_hd_4770/index10.php 

Now, Lusches PSU can deliver a total of 350 W on the 12V rails... so, roughly 300% more. That is a substantial overhead. So much in fact that I doubt the PSU will even run in the load range where it is most efficient.

But I agree, the 9600GT fits his CPU better, unless if he wants to maximise AA and AF settings.

BTW, did you know that the 9600GT consumes on average only 10% less power then the 9800GT? So if a 9800GT is a no-go, the 9600GT is too.   

Bad numbers, the average draw on a 9800GT is about 230W under load. Easy to find a number of sources but here is one...


The other question is actually determining what is available on the 12V rail. According to the label on the PS 145W is earmarked for the lower voltage lines. As a general rule you cant use that wattage in calculating continuous power for the 12V rails, so your already at 305W. While the PS may very well push the full 450W it generally will not provide continuous clean power under load beyond its efficiency rating....so that leaves you with roughly 214W of clean power....not enough.

[humble]

Thought I'd add a few comments, others can offer additional insight or correction as they see fit. PS's are squirrelly things....

The normal wattage you see is the rated "output". So a 500W supply is supposed to pump out that wattage while drawing more AC in correlation to its efficiency rating. Unless specifically listed as continuous output the vast majority of power supplies will provide less continuous power (even the good ones). As a general rule you can expect roughly 90%....so your 500W PS will deliver about 450W of true output under ideal conditions. Additionally not all power is available for each "rail". Many top notch power supplies will list very specific information...


Here is an image of an actual PC power and cooling test page (shipped with PS). so this PS uses its continuous output as listed output not its "peak output" like many others.

This PS also lists its output at a specific operating temp 40C that is realistic in a well built system....many brands test at 25C or even less. As heat builds output drops markedly. note also that this unit operates at 84%+ efficiency. As a general rule the higher the rated efficiency the better the sustained output...especially at normal operating temps in real world use.

So as a general benchmark you can take the rated output and multiply by rated efficiency and then subtract any dedicated output for the low voltage rails to get a reasonably reliable worst case output for your 12v rails. Of course best case you've got a PS from a company like the one above that will eliminate a lot of the guess work. Just realize that even a PS like this will suffer significant degradation in output if the operating temps are 50C instead of 40C....thats why guys like TD make sure that they have a clean build, excellent cooling...and enough pop in the PS to start with.

[Masherbrum]

Well, do you guys actually know what a 9800GT consumes in RL, not on "paper"?

As a worst case scenario, which would be running Furmark, a 8800GT (the 9800GT is essentially the same card) draws about 112 W.

> http://ht4u.net/reviews/2009/sapphire_radeon_hd_4770/index10.php 

Now, Lusches PSU can deliver a total of 350 W on the 12V rails... so, roughly 300% more. That is a substantial overhead. So much in fact that I doubt the PSU will even run in the load range where it is most efficient.

But I agree, the 9600GT fits his CPU better, unless if he wants to maximise AA and AF settings.

BTW, did you know that the 9600GT consumes on average only 10% less power then the 9800GT? So if a 9800GT is a no-go, the 9600GT is too.   

You've become the next "TheZohan".    My figures are from eVGA's website and Leadtek's.    If you paid attention to his post, he has the 450 watt PSU, hence the large "X" crossing out the two other configurations.   

The use of your "giggling smilie" only further accentuates negligence of "given facts" that get "tossed aside".   

Lusche, the 9600GT is the optimum card for your need.   

[Boozeman]

OK, one after another:

Humble: Your graphs look like the consumption of the WHOLE system, not just the Video card. Most test sites are mesuring consumption for the whole system. You need some special gear to directly check the video card consumption, thus not many sites do this drill.

BaldEagle: Please check some hardware review sites and compare the power consumtion of the 2 cards. You will find they are rather closely matched. This is why you have to take manufactueres recopmmendations with a grain of salt. They tend not only to be ultra cornservative, but also take some worst case scenarios into account, that are quite unlikely but may happen. For instance, there may be someone who builds an ultra hardcore cutting edge computer with the finest and most demanding compnents, but puts in a low performance video card. In that case you need a stronger PSU as compared to the same card in a low performance budget system. I mean, no one is retarded enough to actually do this, but as long as the possibillity remains, the manufacturer must take it into account.

Masherburn:
Oh please, no need to get personal. Unlike Zohan, I have facts to back my claim up.  I'm very aware that he has a 450 W PSU with plenty of oompf on the 12V Rails. As for manufacturer recommendations, see above.

 

[Skuzzy]

You do not need special equipment to measure current load and/or drain.  A Watt meter in the hands of someone who knows what they are doing is all you need.  If you can measure the total load, you can measure the individual loads as well, if you know what you are doing.

[humble]

The numbers are for the card itself, they match up almost exactly with the 19@ load out +/-. The issue is really in the benchmarking. A game like AH will put a lot of stress on both the CPU and GPU, not only will the load be high but the sustained operating temps will be at the systems high end. Flying in an event like FSO where your placing a high constant demand on the system for up to two hours may exceed another benchmark by a wide margin. I lean toward the 24-26@ mark over the 19@ for that reason. In addition the heat buildup may lower actual output for the PS.

IMO your under estimating the power consumption of the card and over estimating the reliability and sustained output of the PS. This is a bad combination....
  I'm very aware that he has a 450 W PSU with plenty of oompf on the 12V Rails   

IMO he has relatively marginal "oomph" on the 12V rail. His max amps under ideal conditions is 25@ (305W). Using a very conservative 90% continuous power we have 274W (23@). That is best case and hardly qualifies as "plenty" IMO. However if we have a less efficient unit that measures at lower then true operating temps then you could see sustained output of less then 19@'s total on the 12V rail.

I've run cards on lower output PS's fairly often and as long as you know your way around a multimeter your fine....but I also have found circumstances where you have a lot less "oomph" then the label says...

[Masherbrum]

Masherburn:
Oh please, no need to get personal. Unlike Zohan, I have facts to back my claim up.  I'm very aware that he has a 450 W PSU with plenty of oompf on the 12V Rails. As for manufacturer recommendations, see above.

 

Who's "Masherburn"?   Good Luck in the Future.   I haven't even begun to get "personal".   However, I feel sorry for anyone who will take your "expertise"/"advice".   

You typed "350", not I.    No 9800GT that I am aware of is good to go with "350 watts" as you previously stated. 

[Boozeman]

Humble, check these numbers for reference:

http://ht4u.net/reviews/2008/sapphire_radeon_hd4850/index7.php

http://www.computerbase.de/artikel/hardware/grafikkarten/2008/test_nvidia_geforce_9500_gt_9800_gt/27/#abschnitt_leistungsaufnahme

Similar numbers but clearly statded for the whole system.

In fact, I checked hardware canucks and found this line in thier power consumption introduction:

For this test we hooked up our power supply to a UPM power meter that will log the power consumption of the whole system twice every second.

> http://www.hardwarecanucks.com/forum/hardware-canucks-reviews/17320-ati-radeon-hd-4770-512mb-gddr5-review-19.html

This should make it clear.

[Boozeman]

Who's "Masherburn"?   Good Luck in the Future.   I haven't even begun to get "personal".   However, I feel sorry for anyone who will take your "expertise"/"advice".   

You typed "350", not I.    No 9800GT that I am aware of is good to go with "350 watts" as you previously stated. 

Sorry for the typo. No pun intendend. It's quite late over here, so forgive my sleepiness.

But back to the "meat". If you can show me a test from a hardware review site that indicates that 350 W on the 12V rails alone will not handle a 9800GT, I may consoder you as back in the game. Good luck.