So there is this new GTX 570 sitting...

[denver69 0]

over at Microcenter for $350. The price has dropped allot since I built my computer. I know I have posted about this before but I just have this nagging feeling that I am bottlenecking at my video card. I am now overclocked at 4.8 and very stable. I notice the things that slow me down seem to be video card related but I would like to hear others input on that because I am just not sure. I am flying into Vancouver (Orbx) with 1024 rex clouds in the NGX and am on approach am getting not quite loaded texture scenery, landclass textures and mesh(mega blurries). I notice with the Orbx stuff it is hard on my system. The autogen really hit hard in this area and the low fps and stutters are directly related to building in autogen. Reducing to dense really helps the most in this case. So I am running 8sq AA and no TAA. I cant stand the cockpit of the NGX on anything less than 8sq AA so I may need a better card.

 

How many think I am bottlenecked at the GPU? Just trying to justify the upgrade. I would hate to do it and find out it did not really help at all. For this flight I removed the Bufferpools section and really did not notice any difference. It comes a question of can the card draw fast enough as the CPU wants it to, I do not think it can.

 

Also this new 570 has 2.5gb of video memory as well. I notice in EVGA precision when I have it running during FSX the memory is allways maxed at the 1gb limit the whole time. So I think 2.5gb will help allot.

 

Thanks!

[Ben Cap 89]

 

How many think I am bottlenecked at the GPU?

 

You are very GPU bottle-necked. I'd recommend paying 380 bucks for a bit more horsepower, the GTX580.

[Barnstormer1 0]

Just paid 289.00 for an EVGA 570. Waiting for my broken 580 to be RMA! Seen very little difference between the 580 and the current 570.

[Ben Cap 89]

Just paid 289.00 for an EVGA 570. Waiting for my broken 580 to be RMA! Seen very little difference between the 580 and the current 570.

 

Well, when you start to run SGSS, there is a notable difference in the 580 vs any other card.

[denver69 0]

Yes, I was a little torn between 570 and 580. From what I gather the 570 is just fine for a 1080 monitor such as mine even with SGSS at higher settings. However one would need the 580 to do it at 1200??. A monitor upgrade is a consideration in a couple years due to the fact that I just got this one a couple of months ago. 27" at 1080 was a big step. A 30" at 1200 is a while down the road for me.

 

That is a good deal for that 580 though. Worth the extra few dollars for sure. I wonder would a 570 with 2.5g of ram vs. a 580 with 1.5gb and similar clock speeds be better than the other with FSX. I notice video ram with my card is maxed the whole time during a sim sesh.

[Ben Cap 89]

Yes, I was a little torn between 570 and 580. From what I gather the 570 is just fine for a 1080 monitor such as mine even with SGSS at higher settings. However one would need the 580 to do it at 1200??. A monitor upgrade is a consideration in a couple years due to the fact that I just got this one a couple of months ago. 27" at 1080 was a big step. A 30" at 1200 is a while down the road for me.

 

That is a good deal for that 580 though. Worth the extra few dollars for sure. I wonder would a 570 with 2.5g of ram vs. a 580 with 1.5gb and similar clock speeds be better than the other with FSX. I notice video ram with my card is maxed the whole time during a sim sesh.

 

Get the card I reccomended and OC it. The performance will be about 15-20% faster than a default clocked 570.

[Kosta 122]

nagging feeling that I am bottlenecking at my video card

 

You don't need a feeling. You run the GPU-Z and see if you are.

 

And hold on about these talks 570 vs 580...

 

570 is a great card. Almost on par with 580 according to some reports here and comparisons.

 

Notable or not notable difference. Yes, each card 560, 570, 580 is faster than another. In benchmarks you are gonna see differences, otherwise no reason for one to cost more than another!

The fact is, running 580 is going to give you the best. But ONLY if you really use it. Are you ready to pay for the BMW? Pay if you are and enjoy it.

 

I would refrain hard from saying notable, as for many simmers going from 560ti to 580 is not going to be notable. But it is going to give you an edge, and at some situations even more than just an edge.

[denver69 0]

Word Not Allowed,

 

I said a feeling because of my distrust of what I am seeing in EVGA precision. During the bog down phases in FSX I am seeing only 70% with ocasional spikes upward but rarely 100% for any length of time. However clockspeeds and RAM are maxed. But the GPU usage monitor I would expect to be firewalled at 100% as I get "video card" related slowdowns, slow texture/autgen loading and skips, hesitations and stutters. I just do not believe 4.8ghz processor cant be better than this so that is my feeling because I do not know for sure. I hope it is simple matter of upgrading the card and I will then probably get the 580 as I am currently running fair levels of AA and would like to do better while some other sliders like autogen go to max vs. one step down.

 

Am I greenhornin it to think that the GPU usage meter is telling the truth?

[denver69 0]

I just did another test with GPUZ and performance manager up for the CPU load. flting over UTX memphis with REX clouds and GEX textures with autogen on 2nd from the top, most sliders maxed or close to it and no bloom of course.

 

GPUZ highes GPU load was averaging around 66%.

Memory used is about 600mb (half of whats on the card)

Memory controller load never got above 17%

 

The CPU is a very different story.

 

Core 0 = 20%

Core 1 = 90-100%

Core 2 = 90-100%

Core 3 = 90-100%

 

Mostly at 100% with brief downspikes to 90.

 

This is what tells me I may get more with more overclocking vs. an upgrade of the card. I am just afraid to get a new card and still have the same deal more or less. The numbers seem to indicate I am limited by the CPU. Fore some reason the way it is slow to draw under load and the way it is doing it tells me it may be the card as well and by getting a new card I may UNLEASH the CPU. Don't laugh now.

 

I am running Affinitymask=14

[Kosta 122]

Am I greenhornin it to think that the GPU usage meter is telling the truth?

 

Well, understand it like this: if you are not 100%-ing the card, then you are going to see a some boost between those cards. Probably not as much as you would like. It will allow you though to run better AA at the same expense of the framerate loss.

 

Testing doesn't happen in one single situation. You have to make different scenarios, with high and low IQ, more cloud layers, addons sceneries, heavy aircraft etc etc.

You can't just run one test and say it loads so much... but in this particular test you were CPU limited, so much is clear.

[Tabs 2,790]

"Bogging down" that's resulting from CPU saturation isn't going to show up on a GPU usage meter... EVGA Precision is accurate.

[denver69 0]

"Bogging down" that's resulting from CPU saturation isn't going to show up on a GPU usage meter... EVGA Precision is accurate.

 

I was not looking for that on the GPU meter. Yes I would see the GPU load on EVGA and the CPU load on the perf manager. I had both up at the same time. The CPU's last 3 cores were maxed and the first core(last) was not. The GPU was around 70%. I guess I answered my own question here on where the bottleneck was. I have had several people tell me that GPU usage meter is not all that acurate but I do not have anything concrete to base it on.

 

More importaintly I asked myself why in the heck do I have an Affinity Mask in my config file as I watch the results of this on the performance manager for the CPU. It must have slipped in their somehow :Nerd: . So I deleted it and got a nice 25% boost in performance! This is what spurred this post, I could not figure out why my performance had decreased so much over a few months. I think a card will be good but that is not where the bottleneck was, It was the Affinity Mask. I do not even remember putting it in there.

 

I just cant believe how much that was holding my CPU back. Now all cores are being used and total load has dropped, scenery builds in normally now and fps is up by 25%!

 

Thanks