Issues with 650 GTX Ti & mobo alignment

[azman90]

I just built a new rig with an ASRock H77 Micro-ATX and was temporarily forced to use my old full tower case. The case is a Thermaltake Soprano-- about 5 years old.

Everything worked out fine until I went to plug in my graphics card (GeForce 650 GTX Ti)-- the mobo slot didn't line up with the case ports!

It was off by maybe 1/4 inch or so- I could maybe make it work, but it would put the card at an awful bind!

Thankfully the mobo has a second PCI-E 3.0 slot-- it doesn't say anything about SLI, but it does have a Crossfire stamp.
I put the card in there- it's working well- but I'm scoring 30-40% lower on Passmark Performance Test than the 650 GTX Ti baseline.

Everything else-- ram, hdd, cpu is performing at or above the relative baselines... I've already installed the latest 64-bit Windows 8 drivers.

--I have a new Micro ATX case coming in a week or so, I hope that those ports align! 4 Questions:

1) Have you ever heard of the ports not aligning? Is this a defective mobo, or an outdated case, or maybe I'm dumb?

2) Is there a performance difference between the top and bottom PCI-E slots?

3) Is there a performance difference with DirectX on Windows 8 64-bit as opposed to say Windows 7?

4) Maybe my card is defective? How would I really know here? Maybe it's a power issue? I have a 400w psu?

Edit: it's an EVGA and the 1GB model in case that matters:

5) Would the 1GB cause the massive performance drop?

 

[merandos]

There is no noticeable performance difference between both slots, you are talking about maybe less than 1% difference, and not even in most cases it happens...

Get the GPU-Z and see if the card is running on its maximum Core clock underload...

As for DirectX i really don't know if there is a difference but am pretty sure it won't be lower on windows 8, and even if it is lower, it won't be by much....

And maybe the baseline passmark test of the gtx 650 ti is performed on complete high-end system... and that causes the performance difference


Tell us you resolution and what FPS you get on different games ( provide games titles and settings please )

 

[azman90]

I'll get to the games soon-- preliminary results with GPU-Z show full clock and memory when hitting the tests; however, memory used didn't max out, so it's not a 2GB vs. 1GB issue.

The average is still 2719, and mine is 2081. I'm using the Passmark Performance Test-- the worst FPS was in the island 3D mode with the fireballs-- I was getting around 20 fps. I scored well relative to the Radeon HD 6950 in DX11 and DirectCompute (110%), but got SMOKED in DX9 complex (42%) and DX10 (41%).

I noticed the in the last 5 baselines for the GPU, the lowest was 1682 (next lowest was 2438).

Just had an idea-- I have an EVGA vanilla-- perhaps I'm competing against the superclocked models and custom overclocks?

I have a mid-range CPU i5-3570, but it's still relatively high on the charts, and I have 8GB DDR 1600 Hyper-X, so I doubt that I'm being excessively bottle necked.

Onto a few games. Thanks for your rapid/friendly response.

My main issue here is the mobo/case, I think I'm just being a bit anal about the tests. If I wanted to smoke the benchmarks, I should have spent more than $700 on the build.

 

[merandos]

Maybe.....but don't do overclocking or smthin...just try games and that would be a sufficient proof about how exactly your card is running....

 

[azman90]

I'm installing the Crysis Triple Pack & AC3, so I'll get back with fps for those soon. I haven't been in the gaming world for a long-time (5 years since my last build), so right now I have my CS: Source and Metro: 2033 experience.

I ran both games with everything cranked to the max @ 1920X1200. This includes full anti-aliasing, HDR, bloom, etc.

On CS: Source I got around 120-150 fps with an alltime (very brief) low of 70 fps during a lot of smoke/fire.

On Metro I ranged 14-22 fps, except for one big combat scene where I hit 10 fps.

The benchmark test of Metro had the FPS at 33 avg... It's too early to tell yet, Crysis should be a better clue maybe AC3 also?

 

[merandos]

Everything maxed out on 1920x1200 metro 2033 on gtx 650 ti and avg of 33 FPS, that is just amazing and cool, the metro 2033 struggle most of the cards, and on that resolution i though the 650 ti would be even slower, at that reso with most games you need to lower the AA alil bit..because you are really running a high resolution on a mid-range card

 

[azman90]

Getting 20-22 with a low of 15 on Crysis with 1900X1080 (no 1200?) fully maxed.

I'm assuming Warhead and Crysis: 2 use the same engine/ similar results?

I noticed that it's running DX10, perhaps Crysis: 2 will use DX11?

---

Crysis is almost 5 years old, so I was hoping that I could run it, but you're right about the pixel difference at 1920*1200 (2.3M) vs. say 1280X1024 (1.3M).

I'm still getting only 60% of the benchmarks for this card, but maybe they're done with 16+ GB of ram and i7 overclocked, etc.

 

[merandos]

Yes thats what i've told you, it might be tested on a complete high-end system...

 

[simmons33]

Only other thing I could think of would be something to do with the slot. I mean, I dont think their should be any difference. Is it possible that 16x speed slot is running at 8x? Any bandwidith options in your MB?

Could it be a memory interface issue? Only 128Bit

 

[azman90]

Only other thing I could think of would be something to do with the slot. I mean, I dont think their should be any difference. Is it possible that 16x speed slot is running at 8x? Any bandwidith options in your MB?

Could it be a memory interface issue? Only 128Bit

BINGO! BINGO!

Wow I'm dumb- I'm running on the ASRock H77 -- the top slot is PCI E 3.0 and the bottom is PCI E 2.0!

New case arrives tomorrow afternoon, so hopefully the ports line up and I'll get in the top slot!

 

[merandos]

There is no difference between PCI E 2 and 3....told you it is something like 1% difference, and anyway in order to use the PCIE 3.0 you need the CPU, Motherboard, GPU supports PCIE 3.0...so unless you have the 3rd Gen intel core processors you won't get use of the PCIE 3.0 even if your Card and motherboard supports it

 

[azman90]

i5-3570 is 3rd generation and supports pci express 3.0, but I googled around and it looks like you are right, pretty much no advantage to either 2gb of memory or pciE3.

I was hoping for an easy performance fix and the different slot seemed to be it

I didn't think my cpu or ram would be a bottleneck, if anything I thought they were stronger relative to the gpu.

I didn't see the 3570 as holding me back by 30%+...
but then again, I probably shouldn't place too much faith in the benchmark #s.

Is there a utility that tests bandwith and processes, etc vs the cards stated abilities?

 

[azman90]

New case came in, I installed all the parts and I scored a 2709 vs 2081 before.

The ONLY change was moving from PCI 2.0 to PCI 3.0.

Are the testers wrong/ not using the right combos, or does my motherboard have a PCI E 2.0 defect?

Either way-- I'm happy as a clam now-- thanks for the time & help!!

 

[azman90]

Best answer selected by azman90.