Yup as you have noted you are entirely bottlenecked by the CPU especially at that clock rate. Its easy to test this as you should get more or less the same fps at a higher res/max graphics (minus SS) as on low graphics settings.
Apart from the fact you have a locked multi CPU, the p43 mobo isn’t ideal for overclocking either and anything more than ~410fsb tends to push things on those boards. Did you try slacking the ram timings some, or alternatively ratio the ram to stock just to get the cpu/fsb stable first.
Either way though it will make a minimal difference and you really need a upgrade to a newer architecture to take advantage of a newer card
Those temps look high for stock, considering the max temp in the amd spec is 61C. Likely not the issue anyway, I would try and reproduce this out of game by running prime&3dmark on a loop at the same time.
Personally I would multi the PSU, if software is in order (inc drivers) its likely to be either temps, memory or PSU. Assuming you ran memtest for long enough (usually this is 10 passes) and all temps are within range this leaves PSU as a suspect which is why I asked the model. Also, it could be worth running the game from another drive if you haven’t already to rule out the drive.
It’s VERY unlikely gw2 is the root cause of this, its far more likely you had a hardware failure or degradation which unluckily coincided near a major patch date. There are always bound to be others saying they have ‘crashing’ on a tech support forum of which many will be because caused by a multitude of things, of which in terms of overall player-base is minuscule. It’s up to you to ignore this of course but I genuinely think you will be waiting forever for a patch because I don’t think its caused by the software in question.
There are no true 120hz TVs so I wouldn’t bother with them, as Stormcrow suggests if you are going the TV route you really need one which will support 4-4-4 chroma without any hacks (e.g. EDID OR).
Also the above post isn’t really relevant, you mainly use 120hz for the better response/lower input lag anyway and the only thing that works significantly better at 120fps is lightboost. If you are into competitive gaming get a BenQ XL2411T which does 144hz and lightboost, if more casual a TV 4-4-4 which is capable of reporting edid properly for PCs over hdmi.
This has been present since release, it affects all independent of hardware. As you have noted it occurs when too many sounds are played and the fact anet still hasn’t fixed it after 6 months is pretty pathetic. A workaround is to move the sound quality slider down one notch.
Not played this game in months but occasionally check in to see if they have done anything about the state of the engine. Latest patch update I checked said: “Bag Slots and Bank Tabs are 20% off in the Gem Store through February 26!” still got their priorities sorted I see.
Indeed, my defense against doa pixels is distance selling regulations in the UK, not sure if there is comparable in other countries. Obviously doesn’t protect you after that but at least wouldn’t start on a minus
Hmm yeah reduced would probably have been a better word, stopped playing this months ago so am a bit out of touch.
Of course generally speaking what you say is true, but in cases where the skill is roughly equal the player with with less lag/more smoothness will have a better ability to hitscan more accurately and thus gain a advantage. When I used to play on 56k against lpbs I had to adjust playstyle because it was literally impossible to outaim, comparable but to a lesser degree.
Kinda going off-topic anyways now, my recommendation to the OP would be the benq xl2411t if you play twitch shooters competitively otherwise get a IPS like jazhara stated which would provide much better image quality.
Well I agree with some of your points, by response times I did mean the overall response being hz/input lag/other factors combined rather than the often meaningless G2G manufacs quote
I wouldn’t dispute a good IPS with low input lag being acceptable in many scenarios but the only one I found comparable to a 120hz TN is the Catleap clocked at 120hz. Anything lower and its just not smooth enough for games like QW/CPMA without the higher refresh if you play at a high enough level and anyone on somewhere like ESR will say the same. Obviously as quality goes there is no question who wins there! Also I would concede the amount of games that have these requirements these days is very small.
I can honestly tell you I could identify almost immediately in any game if vsync is on or off assuming it doesn’t have artificial lag introduced (for example GW2 before smoothing was removed). TB improves things of course but its still noticeable by a long stretch.
idd each to their own I guess, would rather put up with a minimal amount of tearing than the horrible input lag but that’s me. I know people who don’t even notice the difference in response between vsync on or off :o
For gaming I couldn’t personally use anything other than 120hz, coming from 200hz CRTs it took me (and most others in the quake scene for instance) a long time to switch over because of the horrendous input lag/response time issues of TFTs.
As previously mentioned, it is of most benefit in fast paced games although you will notice the difference immediately even from mouse movement on the desktop. What I find baffling is people who enable vsync on a 120hz TN thus introducing more lag negating the main benefit of the tech…
I agree with the above, you could try running “bcdedit /set IncreaseUserVa 3072” from the command prompt on the desktop, restart after applying. This should eliminate the out of memory errors although moving to 64-bit is still a better solution long term.
Could you further explain the type of crash? Is it CTD, bsod, restart or freeze+looping sound for instance? Also the above^^ with the addition that drivers recognize and throttle some of the stress testing benchmarks.
Hmm that’s harder to narrow down then, could indeed be a software issue. If it continues after a fresh install I would do the 3dmark stress test and you could also try removing the hdmi audio drivers as these have conflicted and caused this issue for some people.
About how long? I can leave it benchmarking while I sleep so that’s a good 8hours. And what should I look out for? Is this like EVGA Precision where it detects artifacts/visual glitches and keeps them in track while benchmarking?
That would be fine, mainly testing stability of the overall system/power so it will either be running when you come back or will have crashed/rebooted. I think it’s very unlikely it is the card (artifacting etc) being the second one.
Could you further explain the actual BS problem, can you close out of the application when this occurs or do you have to reboot? Does the monitor lose signal?
I would advise you leave 3dmark11 on a loop for several hours to try and reproduce the instability. Furmark is power/fps throttled these days so isn’t reliable enough.
That sounds like either a temp or power issue because reducing the clocks reduces the power consumption and so does disabling HT. Likewise, disabling HT lowers temps in many situations. Alternatively it could be a voltage problem but that is extremely unlikely if you are not overclocked and using auto settings.
This will not necessarily show up in stress testing because running prime95 doesn’t use as much power as GW2, it only uses the CPU. Likewise programs like occt and furmark will also consume less power because they are throttled at driver level.
I cannot comment further not knowing the specs though.
The most common causes of this are overheating, memory and power. It seems you have eliminated overheating for the CPU but it would still be worth checking GPU temps whilst in-game.
Memory – It’s worth checking through your BIOS at the memory settings and running several loops of Memtest. As you are running two sets of ram I notice they have:
a). different speeds
b). different timings
c). different voltages
This is a bit iffy, as you will have to run one set of modules above the stock voltage although no doubt this will be fine but I havn’t checked the full specs and how the corsairs are at 1.65v or alternatively if the g-skill can do 1600 at 1.5v. So for testing purposes you are looking for 1600mhz, 9-9-9-27, 1.65v but will need to research your options here. Also running a full set of modules on some boards can cause issues due to the stress put on the memory controller.
If the memory checks out, it’s likely to be the PSU which as you have noted isn’t too good. It’s ~430w when new with low amps which will only get worse the more it’s used.
Kirito explains it well, I would probably leave the voltage settings at auto at this point. I only suggested it to attempt to lower the temps but its probably better to look into sorting out the cooling.
Even so, I think the problem still likely lies with the PSU. Your question about what can happen in this case, well I have had the following as a result of a shot PSU:
- Crash to desktop
- System freeze (commonly accompanied by a buuurrrr sound), needing a manual system restart or shutdown.
- BSOD
- System restarts automatically
- System shutdown automatically
So at this point you really need to clean the cpu die/block and reapply some good thermal paste at the least and look into testing a new PSU when possible.
- Overheating (GPU/CPU)
- Faulty memory/gpu
- Power stability issues.
From what you have said so far though temps look good, so I doubt its an overheating problem.
For Prime95 you can test overall memory/cpu stability by going to Options>Torture test>Blend. For it to be any use you really need to run it for a long time though, in other words 8+ hours. When you wish to stop the test you can go to Test>Stop.
I would probably first download and run memtest on a loop for a few hours, it will just loop until you stop it and report errors. This will at least eliminate a ram problem.
This likely leaves you with either a shot gpu or psu. I couldn’t recommend OCCT/Furmark anymore alone for stability testing due to the driver throttles which ati/nvidia have implemented as you will get more power draw from playing a game than using them. Running both furmark and prime95 at the same time however may be enough to stress the PSU to a suitable level. I would only start to get worried if the gpu temp consistently hits 80c, the max specs for the 570 is 97c.
Ah I forgot to say you could try your old one/spare to test if you have one
That is actually slightly better than the TT so is def worth testing. If it works I would still consider a replacement in the near future but at least you’d be certain before buying. The coolermaster can only really deliver 450w stable as they again lie about its true output.
It depends what your budget is and also If you are overclocking the CPU/GPU. A good 650w will be fine though unless you have a crazy amount of drives and such.
Something like the corsair tx-650v2 is very good and priced midrange. The xfx core edition 650w is identical inside, both are seasonic units so whichever is cheaper in that case. Seasonic s12ii-620 is another. For comparison, these units will actually supply ~53A as rated on the +12v rail.
As far as thermaltake goes the only good units are some of the toughpower over 600w, everything else is bad or medicore.
As others have noted its likely to be temps or psu, I would monitor temps first.
Now unfortunately whilst that PSU is rated at 650w and 44A +12v it doesn’t actually produce anywhere near this. TR2s are baaad models manufactured by hec. It can probably supply about 30-35 amps at most.
If you tried to pull 650w it would literally explode. I would consider rma for better model.
ah ok, probably just due to the stress test then as long as it isn’t happening in real world sounds ok. I think the GPU is fine at 80c although lower would be nice I think its rated upto 105c.
In that case my first go-to would be the PSU but its unfortunately hard to test unless you have a spare or testing equipment. I’m not 100% sure but I think 3dmark/vantage is still unregulated and running that on loop for a while may be able to reproduce whats happening in GW2.
Hope you get this sorted will check in 2moro am off to bed as its 7:30am here
Okay so, the issue is that your computer is either restarting or shutting down after increasing the voltage. Now i’m assuming temps for cpu/gpu are within range so that can be eliminated. The problem with stress tests is that the power draw is being capped by the gpu drivers due to poor quality vrms on many units so isn’t always a reliable psu test.
Whilst the CX-500 does actually pump it’s specs (just about) you will be pushing its limits being 2 years old (~20% degradation) with the overclock on both cpu/gpu – the extra voltage now causing the restarts is a prime symptom.
MMO’s seem to draw the most power from any game in my experience. On one of my rigs recently swtor finished off a 18month old TX650 and that was only with a overclocked cpu/gtx460 and 5drives….I would reconsider checking there.
Stress tests max the components out to stress the card. Would that not mean (especially in Furmark/Prime95) that it would be pulling as much as it can to power it? It’s what I’ve learned to do to test the PSU when I don’t have the supplies or extras on hand.
I did a stress test with both Prime95 and Furmark. I watched the temps climb to about 85 for my CPU and about 80 for my GPU. Freaked me out so I shut the tests off. I am going to watch the temps in GW2. Shouldn’t be the problem as I play much more demanding games… Well if it is its only a couple $$ to buy a couple of fans (or maybe $$$ to buy a liquid system). Didn’t even consider temps as I have done individual tests and played stuff that cripples cards much more powerful than mine. Will update after I test it. Might not be until tomorrow.
Furmark has been regulated (capped) via drivers->hardware regulation for some time now for both ATi & Nvidia, same with OCCT so aren’t very reliable for this purpose. It is possible to disable this at your own risk. http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_580/ (Page 25 direct link wouldn’t work)
Note the difference between furmark locked & unlocked is a staggering 150 Watts.
The CPU temp in particular is very worrying though and likely the cause of the problems, the max temp from AMD specs being 62c irc. Also for future reference, you may repair by adding -repair to the GW2 command line.
Okay so, the issue is that your computer is either restarting or shutting down after increasing the voltage. Now i’m assuming temps for cpu/gpu are within range so that can be eliminated. The problem with stress tests is that the power draw is being capped by the gpu drivers due to poor quality vrms on many units so isn’t always a reliable psu test.
Whilst the CX-500 does actually pump it’s specs (just about) you will be pushing its limits being 2 years old (~20% degradation) with the overclock on both cpu/gpu – the extra voltage now causing the restarts is a prime symptom.
MMO’s seem to draw the most power from any game in my experience. On one of my rigs recently swtor finished off a 18month old TX650 and that was only with a overclocked cpu/gtx460 and 5drives….I would reconsider checking there.
If you’re getting bsod’s it is not a software bug, its either hardware or drivers. How old is the PSU? Also is your cpu overclocked in addition to the gpu?
As kirito notes above, you are cpu bound in those scenarios.
But for my GF visiting Turkey with no medical insurance ; it would have been an i7.
I had to replace crashed Phenomx4 980 system with only a marginal upgrade.
Of course there was wintersday patch during my downtime; so can’t make an accurate comparison between the old and new rigs.
@Kiroto .TY for the links Catalyst settings seemto be worth +5 more FPS in zerg and a steady 60 elsewhere. But for culling, thats at least a bit more playable.
To be fair, the state of the engine is atrocious so you aren’t missing all that much. Hitting 30fps under the best system possible is not really acceptable but i’m not expecting any improvement soon.
In-fact at the rate they have been fixing the vast amount of bugs (or not fixing I should say rather introducing more bugged content) I doubt anything will be significantly improved before an expansion!
The above are good calls, however it would be easier for people to help if you stated your needs (overclocking/no overclocking?) and budget.
Personally I would look into a better quality PSU depending on your budget/specs and its almost a must if you are going to be overclocking the cpu/gpu because that supply becomes unstable when reaching over ~75% capacity.
The above pretty much covers it just a couple of things. The tjmax for that CPU is 100c but it shouldn’t get anywhere near this (wouldn’t really like to go over 80c) so obviously a major problem.
You could try manually lowering the vcore to reduce temps but may have to try a few different values to maintain stability, knowing the vid would help which shows in realtemp/core temp. You could try starting at say 1.00v, by changing the auto setting to manual so the below option becomes available. The auto setting generally speaking over volts a fair amount.
This won’t really fix the core problem though so recommend following SolarNova’s advice above, sounds like the whole loop needs examining/blocks reseating.
You were right, I did look at that review and they do have the wrong label, however according to the image you provided which has the P12N, it is 2x 12v Rails with a total of 43A but indeed that doesn’t tell us anything about the Power Factor those units use.
Yer as i’m sure you are aware most trash brands outright lie on the label, the only way I found out the true amperage is by pulling up the actual specs from the manufacturer SuperFlower and working it out that way.
Sorry I should have been clearer, what I meant by testing is basically swapping out your psu with your friends to test.
Another method is to use a digital multi to monitor the voltages under load of your friends current psu although I wouldn’t recommend this way unless you are confident with it. Can provide more info if needed.
I think solar just meant with regard to the dust can adversely effect temps but if they are fine there should be no issues. Wouldn’t vac though due to static, compressed air would do the trick
Would probably be worth testing your psu if it’s not too much hassle if only to eliminate it (depending on wattage and such). As far as a new PSU goes it would depend on how high you want to go but for the current system a good 600w will do. Some good oems are seasonic, delta, zippy, enermax. The list here gives some good basic information: http://www.eggxpert.com/forums/thread/323050.aspx
I would recommend something like the seasonic s12II-620 but it really would depend how much future proofing you want with capacity and if modular is important.
My numbers were definitely correct nitro, I assume you got the numbers from the JG review which is incorrectly labelled (it’s actually the p14). Even then, no way in a million years is it 74A at combined 12v ratings due to their fantasy specs. If we work it out, the p14 is 50A on 12v (600w/12=50) but it wont even run that stable at full load, it drops below 11.8v on 50amp loads with nothing on the 3v/5v.
The p12 on the other hand has 442w combined on +12v (442/12=36.8A). That’s cutting it very very fine especially if you bare in mind what these units perform like at full load.
TLDR; bad psus – avoid.
Following the advice above and balancing components on the rails would be the best case scenario & testing a spare if possible. Not saying the cause is the psu though it could be any of the above suggested previously.
Definitely worth flashing to the newest bios for sure. As far as that psu goes its pretty bad honestly, runs about 36A on two 12v rails (not factoring capacitor aging) which is very low for a 700w. By comparison a mid-range quality 650w such as the corsair tx-650v2 runs about 53A. Firstly you will need to ensure each pci-e power connector for the graphics card is from a different cable set or more accurately another rail so that it balances the load over both 12v rails.
Obviously worth going through the elimination process with everything else as well.
As you are running stock I would not bother touching the voltage settings at this point. Reason being every chip requires a different voltage to be stable and without extensive testing you will not find this especially as the instability issue could be elsewhere. ‘Auto’ whilst not ideal for o/c circumstances generally overvolts to some extent and therefore its very unlikely to be causing this.
I would look into checking the following:
1). RAM – run loops of memtest to eliminate this and set the timings manually in the BIOS according to your modules. Mainly looking at Cas, R2Cas, TRP & tRAS here.
2). Monitor GPU temps, I note you checked CPU temps are fine in the OP but didn’t mention GPU so unsure if these were checked within range as well?
3). PSU – This is probably my #1 suspect especially as you note this was a pre-built system. Most pre-builds include a terrible PSU to save costs, make/model would help here if possible as nitro mentions above.
The problem is that the max addressable ram is 2GB in 32-bit for a single application under any circumstance, the maximum ram installed is irrelevant in this instance because it simply cannot access it due to no LAA available to 32-bit OS.
The 3GB switch is a legitimate workaround but you do have to be aware of the amount of background kernel processes when using this. I suggest you test using this first but there really isn’t any reason not to install 64-bit as it is free as the keys are identical…
If you are certain temps are fine its likely to be a power issue, be it the actual brick (easy test) or the mobo voltage regulation. If you’re wondering why this wouldn’t show up in OCCT for example, its because some of these programs are throttled at the driver level due to poor quality VRMs/VRM cooling some manufacturers use…
