Benchmark seems VERY low

a-fogle

Distinguished
Feb 16, 2010
15
0
18,510
I am unsure where to post this, but I've seen some other benchmark threads throughout Tom's. Anywho, I recently decided to upgrade my graphics card and power supply (to accomodate the new card) and I did not notice a significant gain in performance. I believe there is an issue within my system, and I would appreciate outside opinions. Here is my setup:

Mobo: Asus M2N32 SLI-Deluxe Wifi
CPU: AMD Athlon 64 x2 6400+ (stock, not overclocked) at 3.21 GHz
RAM: 8GB Corsair XMS2 (5-5-5-18) DDR2 800 (PC2-6400) (although recently it is showing 7.5GB usable, don't know how or why, used to show full 8GB)
HD: Western Digital Caviar 320GB (16MB cache)
Creative Soundblaster X-Fi Titanium
Lite-on DVD-RW Drive
Antec 900 Case
Windows 7 Pro 64-bit

NEW: Graphics: BFG GeForce GTX 275 OC Edition - 896MB @ 448-bit - 648MHz core clock, 1440MHz Shader, 2304MHz effective mem clock
PSU: Antec Earthwatts 750W PSU - 4x+12V rails @ 25A each

OLD: Graphics: XFX GeForce 9600GT - 512MB @ 256-bit - stock settings
PSU: Antec Earthwatts 500W PSU - 2x+12V rails @ 22A each

I have not run 3dMark06 with the old setup, but with the GTX 275 and the new 750W PSU, 3dMark showed me around 7100. This seems AWFULLY low compared to other systems with HALF as much ram, same graphics card, and same or similar CPU. I don't know what I should try and diagnose, but this is quite aggravating. All drivers are up to date, and I never get BSOD, so I do not believe my system is unstable, but perhaps something isn't correct. I do know that the GTX 275 should be supplied with sufficient power with both 6-pins connected as they are.

Any help? Thanks. Maybe people know of comparable systems? Like I said, a mark of 7100 overall seems very low with the graphics card and decently clocked CPU.
 
Normal? Really? I saw people with similar setups, some with less RAM, and getting a mark of 12000+. I tried CODMW2 on single player (because I'm at college and our network is fairly slow), and my FPS was quite low on one of the first levels. On COD4, I used to dip quite low into the 20s on a couple maps with my 9600GT. Same thing with this GTX 275, so I don't get it. I thought about running FRAPS to see what I get in MW2, but I think I'll still be disatisfied. My brother in law just upgraded to a Core i7, overclocked to 3.5GHz (I think), 6GB DDR3 1600 ram, and he has 2xGTS 250 1GB in SLI and he marked at over 21000. I don't see how he got that much performance gain over me. COD4 and MW2 are really the only games I play at the moment, but I feel as if my current setup is more than sufficient to run at max settings with no noticeable video lag.
 
ok exact setting on the 06? 12000+ may be on a lower res or setting and that will affect the score


also, that cpu could very well be bottle necking the 275, which is a powerful card...

i would say maybe try to play a game with different res (eg 800 * 600 vs 1600*1200) and if there is a no or minor drop in performance (even if running at say 10 fps) then it would mean that you are cpu limited and thus this upgrade only solves 1/2 a issue at best.

that ancient thing is a K8 derivative, which is based on the 2004 model from amd... the i7 obviously would have slaughtered it since the core 2s slaughtered it already...

and looking at the compat of that mobo, it seems that you can't even try to run a newer Ph II on it (K10 core)

 
Sorry, I forgot to mention the settings of 3dMark. I ran it with just the default settings (since I'm using the free version) which is like 1280x1024 or w/e. I normally use my 22'' LG (1680x1050 of course).

Saint: Like I mentioned, I haven't run tests to check FPS on MW2. On COD4, sometimes it will dip into the low 20s, and sometimes hover around 20. If you're familiar with the game, on MP, maps Creek and Ambush are the worst, to be exact. I still need to try something like FRAPS for MW2.

To reply overall, I have read a few things that suggested my CPU may be bottlenecking, which was my greatest fear. And yes, I am aware the 6400 is socket AM2 which is now becoming obsolete. I also understand the Core i7 is fairly brand new and quite the dominating CPU, but I just couldn't see my highly clocked CPU getting that low of a score. Perhaps when it all boils down, the CPUs in my series have low cache values. I wasn't trying to say the Athlon 64x2 is anywhere close to a Core i7, it just seemed quite low. I thought about running tests with the 9600GT I still have and seeing what 3dmark outputs.
 
3dmark is much more CPU bound than most games. As a result, your CPU is heavily bottlenecking in 3dmark, even though it's probably a less severe bottleneck in games. You would probably see a performance improvement in games by going to a faster processor, but not as much as 3dmark would imply.

For an example of this, my 4870 quadfire system scores 17k in 3dmark06 with my CPU at 3.33 GHz (i7), while at 4.2GHz, it scores 27k. In games, there's almost no difference.
 
You can't compare the performance of 2 gts250 in SLI against 1 gtx275, also you are comparing your system with a quad-core (i7) and tri-channel. DDR2 against DDR3.

If you can get mor FPS, you need change at least, your CPU.
 
I know you can't compare the i7 to the Athlon 6400+. That's like comparing a Lamborghini to a Honda Civic. I was more comparing what I saw results from similar setups. I'm fairly convinced it's the CPU, just wish I would have known. I'm going to return the new PSU and graphics card because there's no point in keeping them for now. I'm on a tight budget for a while , so I'll just get back the money it took away anyhow. I knew the GTX 275 was a massive card, just never thought my CPU would drag it down so much. But again, I understand it's based off of 5-6+ year-old technology. And also, I realize the 2xGTS 250 totals 2GB ram at 256-bit, and results in a few more stream processors than the 275 has. Hopefully very soon I can build a new system, and probably go with the i7.
 
As I said, 3dmark is far more CPU-bound than games are. Your CPU may well be able to keep up in games better than it does in 3dmark. At least try a couple of actual game benchmarks before becoming convinced that there's no point at all.
 
Well, that score seems more than a bit low. I got a higher score with my two 3850s and my old Athlon X2 5000@ 3.2Ghz. Heck I got a higher score with one 3850 (around 8k with the overclocked CPU). Certainly seems like something is amiss. So

1) Did you uninstall the old drivers for the 9600GT and install newer drivers for the GTX 275?

2) Check to see if the clock speeds on the card are increasing under load rather than staying at their idle speeds?

Personally I would do a clean install to try and eliminate a corrupt XP install or any driver issues as the cause. If you do return parts though you may want to consider simply adding a second 9600GT to run in SLI. An Antec Earthwatts 500W is more than sufficient for that, and in fact was more than sufficient for your system with a GTX275.
 
Megaman: Yes, graphics drivers are up to date (nvidia's latest is 196.21). Also, I have Windows 7, not XP. Also, I used to have 2x9600GT but recently sold one to a relative. The dual 9600GT didn't seem to perform significantly either. This is why I was considering a CPU bottleneck. I never considered running 3dmark on that setup, so I can't revert back. I can, however, run it with the single 9600GT just to compare with only changing graphics. Since you got a higher score with a lower end CPU clocked at the same frequency, I'm quite confused now. Also, I contacted Antec and they recomended a higher PSU than the Earthwatts 500 even for the GTX 260 (I considered this before I bought the 275). However, a friend of mine is somehow running the GTX 285 2GB SC edition on the same EA-500 PSU. So idk... I didn't think the EA-500 would be able to support any of the GTX 200 series because of the amperage they draw. Also, I checked Nvidia System Monitor AND Rivatuner.... Both showed the 275 running at full settings when in-game. I contacted BFG about the dynamic power settings and it is normal. The idea is it reduces wear and tear on the card when not running heavy video applications. Ultimately, I can't really afford to do a re-install of Windows 7, as I am quite busy with school. I will continue some research, but it is puzzling to me how my PC marks lower than your system, megaman.
 
^ Yes I meant Windows 7. Just so use to re-installing XP I suppose 😀. For Vista I keep an image of a fresh install but haven't had to use it much, thank goodness. It could be Win 7 doesn't like your system, so I just don't know. It's true that an Athlon 6400 X2 would bottleneck a GTX275, but it shouldn't be by that much. Like I said, I got a better score with a 5000+ X2 and a single 3850. Then again I did score it in XP so perhaps your choice of Windows 7 has something to do with it. I have most of my benchmarks, and work related stuff on my XP partition while only using my Vista partitions for games and the web ^_^.

Well, see if your scores go up using the 9600GT. If they do, then it's possible that something is wrong with the GTX 275. If not, then it may indicate another problem. Good luck.
 
Okay, so I haven't tried the 9600GT yet... But I downloaded FRAPS to test performance in MW2. FRAPS showed swings anywhere from 20-90 FPS on multiplayer. the 70-90 FPS range occurred in areas of less activity and detail (such as rocks, corners, etc... expected areas). In areas of more wide open detail, buildings, etc., I was averaging between 20-30 FPS. Granted this was on only one map, but I'm pressed for time at the moment. I tried single player, and it was even WORSE (which may support the argument behind CPU bottleneck). On one of the first levels, I was AVERAGING between 11-15 FPS... this is god awful. Remember these all occur with the GTX 275. Something seems terribly wrong..... My friend with the AMD x2 6000+ (model right below my CPU), same Antec 500W Earthwatts, same exact 8GB Corsair RAM, and motherboard directly below mine (M2N32 SLI Deluxe, no wifi).... not sure how he managed basically the same setup..... He is running all of that with the 2GB GTX 285 and he has no complaints about the game. I will ask him tomorrow to see if he will run 3dMark. I'm getting quite frustrated. :fou:
 
Check the res he is running at...

short of that, turn off cool and quiet and see if it's that stupid bug...


also is it heating up too much? what cooling do you have, get coretemp (or since it's old and core temp may not get it) speedfan and check ur temps

 
Sorry, I forgot to mention it. The friend with the GTX 285 uses a 26'' Monitor, so I believe that is 1920x1200 res. Cool n' Quite has been off, as far as I know. I use the Antec 900 case, and it has a top 200mm fan, front 2x120mm fans, rear 120mm fan, and I have a side window mounted 120mm fan. The front fans are intake, the side panel is intake, and the top and rear fans are exhaust. My system usually runs fairly cool. I can check nvidia system monitor and even after gaming CPU temps stay in the 50s or 60s C. I don't know if this is a significant issue, but my motherboard also is not recognizing memory at full capacity. The two DIMM slots (I believe channel A) nearest my CPU will register a full 4GB of ram, but once I install an additional 1 or 2 sticks, to total 6 or 8GB ram, I lose .5GB for no reason. This then has the BIOS registering 5.5 or 7.5GB, instead of full capacity. This wasn't the case a few weeks ago. I sent Asus an inquiry, no reply yet. The only things I have changed are PSU and graphics card, but neither should affect the RAM problem. All power cables from PSU appear to be connected correctly.
 
hmmm weird...

who knows maybe a partially failing something somewhere is causing this

how about taking out 4 GB of ram (only use 4 GB matching ram and set it at the correct timing and voltage)? the imc on board may be getting glitched out over the long use, 50 or 60 c is kinda hot for a K8 chip in my books, what cooler are you using?

 
Right now I am showing 46 and 45 degrees C for Cores 1 and 2, respectively. I use the AMD stock cooler, never went aftermarket. I suppose I could try running at 4GB and see if I get anything different. If not, then RAM shouldn't be an issue.
 
Just ran 3dMark again with 4GB ram, got 7132. RAM does not appear to have any effect, and I suppose neither does any motherboard error I may have. I believe my friend is going to run 3dMark tonight and let me know how he scores with his similar system.
 
make sure he uses the same setting as you, get him to use the demo version with the fixed res and etc and what nots

this is really weird if your systems are really that similar, unless ofc his athlon is a newer cheap K10 rather than a older expensive K8

maybe get him to send you a ss of the benchmark with cpuz open and do it ur self and see the exact things inside...
 
I swapped back to my old 500W PSU and the 9600GT, got a mark of 6122. So... Adding the GTX 275 helped some, but I think you're all right about the CPU bottleneck. That's quite crazy imo. With the GTX 275, I only get an extra 1000 marks. Hardly enough to notice a performance gain.
 
new cpu time... it's chocking that 275

just grab a 955 BE + mobo and ram and you should be good to go

if you want to save some , grab a MSI 790GX-8D that has both DDR2 and DDR3 and a 955 BE

I would try not to get any DDR2 only stuff now, but if you must there are plenty of early Ph IIs that are DDR2, or just buy a new Ph II and drop it into a am2+ ddr2 mobo or a am3 ddr3 mobo
 
Err 120 before newegg ran out of stock, I think the market for it has gotten small since ddr3 = ddr2 in pricing and that am2 is getting phased out and ddr3 + am3 becoming the de facto standard.