need help please...kinda urgent

Toggle sidebar Toggle sidebar
B

bttkcooled

Distinguished
Nov 24, 2006
23
0
18,510
long story short. I build a new system, I kept changing thing to make it better/ faster ..............cliff notes on the bottom if you dont want to read it all.

I started out with

intel d975xbx mobo
pentium D 3.4ghx
4 gig ram
enermax liberty 600w PS
1900xt crossfire set up ( 2 cards )
2x 80gb in raid 0
500gb
dvd
dvd rw

I was bench marking lil under 9000 in 3dmark05. so after researching, reading and talking to others, that was CRAP for a score with my set up. they thought my CPU was bottle necking me. I also was having problems running crossfire on this board. so I traded them for a EVGA 79050 GX2 great card I like it, I read and was told it will bench better

so I go into 3dmark05, run another test with the new card...STILL under 9,000, so now im getting pissed

so ive been running it as is for the past 2 months...no complaints. so I finally ran into so extra cash, so I broke down and went out an got a intel core 2 duo 2.4ghz ( got it for a steal, 300 new )

so I install it HOPING to bring up my score.
.
.
.
.
.
.
.
this time I could BARELY touch 8500

now heres the kicker, I just did a FRESH format reinstall, I have ALL up to date drivers for EVERYTHING



my question to you guys...what could it be????? what could be wrong? from my knowledge, I could be hitting 13000 something on default 3dmark05 settings...or am I expecting to much?

cliff note:

intel d975xbx mobo
Intel core 2 duo 2.4ghz
4 gig ram
7950 GX2 ( 1gb card )
enermax liberty 600w PS
2x 80gb in raid 0
500gb
dvd
dvd rw

TIA
Scott
 
ok...this is a bit too obvious and you've probably thought about it long ago...so just ignore me if its already been dismissed:

Have you very many programs running in the background, virus checkers etc? have you disabled non essential services and apps at boot up?

I'm not sure how high your scores in 3dMark05 should be, but as a matter of interest has it really impacted your real world performance?
 
format/reinstall was done 3 days ago..only background processes are mcafee, AIM, xfire. now those dont even take up much. ive read on sites and from other people with KIND OF the same set up as me, if not as extreme per say as mine, an there hitting 12000 and 13000 for a score

im WAY off that!
 
Your running xfire in the background? Is that the program to run cross-fire? I'm not sure cause i use Nvidia.

Other than that, have you checked your M/B bios settings? I can't remember with my PCI-E board, but is there some setting that sets bandwidth/aperture for the PCI-E slot?
 
xfire is a messaging program for gamers ( to see whos playing an where )

and I havnt messed around with my bios settings for this at all. I havnt done that since AGP crap

but do you think thats the problem?

if so, what type of settings would I change?
 
well..theres an easy way to check...just remove 2 gigs...though I'd doubt that is the problem.

I'll be honest, i'm a little out of my depth with sli topics...but it does seem to me that tweeking your bios might be an option - particularly if your previous set up was crossfire and your new one is nvidia based.

Also try and find what bios version those guys with the 10,000+ scores were using.

I wish you the best with it.
 
well it had NEW bios drivers loaded with the fresh format/install. so no matter what my system had/done before, it shouldnt matter!
 
oh...sorry...didn't realise you'd updated the bios version also. I'm afraid I'm out of ideas as you seem to have covered the usual suspects.
 
maybe you video card is starved for power - make sure that the 12v/5v cable is only plugged into the vid card and nothing else like hdd or fans etc.

i doubt this is the problem thou.

when it POSTs, does it say that dual channel is enabled? even if it isn't i doubt this would affect video performance. but still, i would try it with 2 x 1Gb and make sure dual channel is working.

PS are you running Nvidia 9X.X video series drivers?
 
if you havent already tried this
set all your graphics card settings to default .
Anti Aliasing and Anisotropic filtering to application controlled.
everything else to quality settings and retest.
see if your scores are in line.
 
How this for an answer DONT PLACE ALL YOUR HOPES IN A BENCHMARK.

Just because a benchmark says 9000(BTW i got a single core P4 and 7900GT and in 05 i scratch 7055) how does your games run i would assume awsome!
 
honestly games run great ( I play league for battlefield 2, where as im sponsored by logitech and mushkin! ) I get around 80-140fps on 1600x 1200

but I read all these things on the video card, all say 12-13000, how are THEY getting those scores? an im not? just curious
 
I was thinking HDD too, BUT if it were that, while 3dmark is running, it would be choppy, BUT its not!!! so I kinda ruled that out!

I will run a memtest tomorrow ( I have all the disk at work )

I know for a fact the psu is fine, cpu is fine...its either mem, mobo, video card. but I dont know which..

why wont the intel board run at 16x if it SAYS 16x? I hve 3 pci-e slots, the top most one ( by the cpu ) is the ONLY one that is 16x and thats where my card is!!!

how do I get the memory module info? and how should I get you all my info so you can help me?

verndewd, do you have AIM? if so, IM me at bttkcooled please

TIA
 
:?: Can you link to the people and / or machine specs / ORB results of those whom claim you should be getting 9,000+ 3DMarks, and indicate which version(s) to which they are referencing ?

The only "http://" search hits on this page are in post signatures, which means no-one has linked to any proof. (There is no yardstick, of known true score, to compare against).



I don't think the problem is the system, I think it is the way 'some' (ignorant, but no-one here) people may be comparing your hardware to other gear with significant overclock scores while you are not overclocking or performing extensive* large tweaks.

* - (read: output video during test is garbled mess of vectors and corrupt textures)

The people that score highly in 3DMark don't always run without massive rendering issues, they just barely finish the benchmark to generate a score, even if the test is mostly garbled (3DMark does not perform integrity checking on rendered output, which is why the entire Futuremark ORB database has slowly gone to **** over the years).

Try running a Half-Life 2: Lost Coast benchmark at high settings, and compare it to a known good value from TomsHardware, or other reputable sources. (Not random smacktards who overclock hardware to destruction, such as 'some' people on the ORB forums).

I've got several documents (not on hand atm, but around would take 3+ days for me to find them) that explain 'flaws' with the 3DMark methodology, the database skew on scores, lack of 'official reference' scores for 'standard' hardware, and various other problems a so called "Professional" online society should implement.

I'm not saying 3DMark / Futuremark is bad, just that ORB results are far from perfect for some of the above reasons, and many I haven't mentioned.
 
I am well aware of this...

But is he comparing his PC to 'similar' ones that have been overclocked +25% on all parts, with extended timings on GDDR3/4 Video RAM, volt-modded, and tests that don't actually render what 'normal' people want when testing with 3DMark, just to get the 'highest possible score'...

OR

Is he comparing to PCs that actually run the 3DMark tests normally, and render normal output. Without mipmap tweaking, and various other 'performance enhancements' for 3DMark which drop quality like a rock in the name of getting a higher score (inconsistent testing / comparisons are generally pointless, except from an experimentation perspective - and he seeks an honest comparison, not experimentation using non standard driver 'optimizations').
 
I was bench marking lil under 9000 in 3dmark05. so after researching, reading and talking to others, that was CRAP for a score with my set up. they thought my CPU was bottle necking me. I also was having problems running crossfire on this board. so I traded them for a EVGA 79050 GX2 great card I like it, I read and was told it will bench better

so I go into 3dmark05, run another test with the new card...STILL under 9,000, so now im getting pissed

so ive been running it as is for the past 2 months...no complaints. so I finally ran into so extra cash, so I broke down and went out an got a intel core 2 duo 2.4ghz ( got it for a steal, 300 new )
Click to expand...

Well, all I can say is that your score is kinda low...
I scored 12850 with a single X1900XT (system listed in sig)
However, my cpu was @ 3200Mhz.
I think I got 11.000 something with everything on stock settings.

I hope you solve the problem.
 
I started out with

intel d975xbx mobo
pentium D 3.4ghx
4 gig ram
enermax liberty 600w PS
1900xt crossfire set up ( 2 cards ) - [Very nice setup, the latest Catalyst drivers on i975 would've rocked on this more than your current card]
2x 80gb in raid 0 - [- Also very nice]
500gb
dvd
dvd rw

...they thought my CPU was bottle necking me. I also was having problems running crossfire on this board. So I traded them for a EVGA 79050 GX2 (great card I like it), I read and was told it will bench better.

So I go into 3dmark05, run another test with the new card... STILL under 9,000, so now im getting pissed

... So I broke down and went out an got a intel core 2 duo 2.4ghz ( got it for a steal, 300 new ) - [Good move]

so I install it HOPING to bring up my score.
.

.
this time I could BARELY touch 8500 - [What were you getting prior to the CPU upgrade ?]

now heres the kicker, I just did a FRESH format reinstall, I have ALL up to date drivers for EVERYTHING

my question to you guys...what could it be????? what could be wrong? from my knowledge, I could be hitting 13000 something on default 3dmark05 settings...or am I expecting to much?

cliff note:

intel d975xbx mobo
Intel core 2 duo 2.4ghz
4 gig ram - [Is it 2 x 2 GB, or 4 x 1 GB ?]
7950 GX2 ( 1gb card ) [What GPU/ GDDR3 speeds are you using ?]
Click to expand...

Questions:

Is every score you exchange with others using 3DMark06, or 3DMark05 ?
Are the scores for the same resolutions (1024 x 768, 1280 x 1024) ?
What score were you getting prior to the CPU upgrade ?
What score were you getting prior to the Video Cards been replaced ?

Can you complete a 'score matrix' for your system (shows history in a table or list easily) ?:
Pentium D + 2 x Crossfire Radeon X1900 series: ?,???
Pentium D + 1 x GeForce 7950 GX2 series: ?,???
Core 2 Duo + 2 x Crossfire Radeon X1900 series: ?,???
Core 2 Duo + 1 x GeForce 7950 GX2 series: ?,???
Version of 3DMark 05/06 used: ?
Resolution used for comparisons: ?
Is FSAA / Anisotropic filtering forced in driver or in 3DMark, or neither: ?
Is there an audio benchmark in 3DMark 06 ?, and do you have it running via cheap onboard audio ?

Is your RAM setup as 2 x 2 GB, or 4 x 1 GB ? (Can make a difference)
What GPU core / GDDR3 VRAM clock speeds are you using ?

Have you checked SLI Zone for advice on SLI setup / performance issues ? - Since you started with a pair of ATI cards then changed to a 'SLI on a stick' card it may be worth investigating the SLI Zone site at: www.slizone.com

This small font section is basically throwing you in the deep end, and hoping you can swim:
FAQ is here: http://www.slizone.com/page/slizone_faq.html ; but it is heavily watered down on the 'how does it work via PCI-Express, and how do PCI-Express devices map memory address space for accelerated I/O' subject. 😛 :cry: (You'll notice in System Properties that you only have 2.75 to 3.50 GB of memory under Windows for apps, because much is used for accelerated I/O via mapping devices to memory ranges, and it is the address range that is 32-bit. As a result not all the total physical memory can be mapped to it - As you have 4 GB this might reveal the underlying performance problem. (Explanation is well beyond the scope of this thread, but search for some of my posts over the last 1+ years and you'll see the gist of it). It is for these reasons most people only recommend people use 2 - 3 GB systems, as it doesn't 'overlap' with the top end of the range that is traditionally used for accelerated I/O mapping (even back when PCs only had 64 MB + PCI bus this was the case, just SLI threw it a curve-ball since SLI needs a huge range to work).

Device_Manager_Memory_Ranges_101.png


Image is taken from my old (4-way) Opteron 270 + Tyan K8WE (S2895) system. If you can convert hex to decimal, and understand 32 bit = 4294967296 bytes mappable address space (not all of which is mapped to memory).... well... it becomes quite obvious to most people by the time I mention this.

Have you run SiSoft SANDRA to confirm CPU performance, and Memory Throughput / Cache size related tests all check out as expected ? (eg: If Dual-Channel isn't working it will show up within 60 seconds of memory benchmarking in SiSoft SANDRA)

Tip: As you are running 4 GB, well even if not but may flag a BIOS default differently, you might want to ensure that SLI Broadcast Aperture is set to a decent size in the BIOS.

If you complete the score matrix above more people will be willing to help out, each adding their 2c, if not presenting you with a solution
 
Oh, and as you changed from ATI to nVidia have you since re-installed the OS after changing video driver 'brands' ?

Under the most extreme cases this seams to help, esp if you changed the video card brand over after the re-install mentioned above.
 
Personally I think the score he gets it correct.

But if he believes it isn't because of various reasons, and yes the total re-install was bit over the top (traditionalist here, forget driver-cleaner even existed as I use Ghost 2003 to restore OS partition to pre-video driver state), but does anyone have any idea what it could be ?

I was thinking FSAA forced in driver or app, Resolution differences in testing, maybe lack of dual-channel mode being active (which SANDRA can test). Maybe V-Sync is locked on in driver + low refresh rate or something (?).

So far I've only got conjecture to work with because the 'core' questions* have not been answered, or answers not disclosed.

* (a good foundation will lead to a good result)

You think I am going all out here, you should see me in military simulators. :lol:
 
other have pretty well covered things but for my 2 cents
make sure you have the latest driver on a clean install of os
disable all backround programs for test (or better don't install them until after)
check brand of ram and voltage and timing specs and make sure the settings are correct in the bios
as for what you should score i get 9345 in 3dMark 05 and just from the specs you should kick the crap out me (even with my oc)
 
yes, everything was formated and XP pro was installed, so therefore ANY old ATI crap is DONE.


but good news. I did another test with 2 gb ram INSTEAD of the 4gb and ended ALL my processes and got a 12563. so I think I might have found my problem

im going to put the other 2 gig in, END all the processes, and try it again

brb, ill be sure to post my results
 
well I put in the other 2gb to give me a total of 4, ended ALL processes. I got a score of 13214. that seems about right...right?

this is with EVERYTHING at default. NOTHING was overclocked.....yet........so I think I found my problem

my mcfee was KILLING me

thanx guys.

im going to overclock the core 2 and vid card now...be back in a bit.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom