Huge FPS drops On A High End Computer

[SuspectDown]

From what I've known to have an overclock that is considered truly stable you really need to run the prime test . Whilst Linx is great for stress testing temperatures I wouldn't personally rely on it enough over 30 minutes to call my overclock completely stable.

 

Many people would run both though personally if I'm going to run a prime test anyway I am fairly confident it will find anything wrong that linx runs would have anyway. The same cannot be send about running linx over prime.

 

You could argue each to their own but if you are proclaiming the death of Prime testing or having it as a thing of the past you are just ignorant.

 

Linx is more effective than prime, this is fact. Linx was also the first to add AVX extensions, something P95 eventually followed through on. I've had overclocks do fine on P95 and lock up within 60 seconds using Linx w/ AVX.

 

It seems every thread that comes up about PC performance you want to go round and round. I've been on EOCF and XS for over 10 years and have recorded many impressive overclocks as well with cooling like I said ranging from air to phase change. Look me up on those forums: Urbanfox. < 10+ years of internet history for my resume, what's yours?

Edited August 6, 2015 by SuspectDown

[jdi_knght]

I'll take true stable over "I've cranked my overclock too high or not running enough voltage but need an app to tell me it's stable" any day.

 

Ah, good luck on that! I've had overclocked systems that survived days of Prime95, hours of OCCT/Linpack/LinX, but would crash 10-24 hours into their work in a render farm until they were clocked down some.

 

"True stable" is pretty hard to ascertain. If your machine crashes once in a year due to the OC, do you consider it stable? Or stable enough for you? And if "true stable" is the end goal, we should probably be using Xeons with some ECC and avoid the whole overclocking discussion to begin with! There's a reason Intel bins the way they do (well, a couple reasons, but we'll pretend they don't downclock higher-binned chips to meet market demand for the lower end CPU's).

[MeNaCe-NZ]

Linx is more effective than prime, this is fact. Linx was also the first to add AVX extensions, something P95 eventually followed through on. I've had overclocks do fine on P95 and lock up within 60 seconds using Linx w/ AVX.

 

It seems every thread that comes up about PC performance you want to go round and round. I've been on EOCF and XS for over 10 years and have recorded many impressive overclocks as well with cooling like I said ranging from air to phase change. Look me up on those forums: Urbanfox. < 10+ years of internet history for my resume, what's yours?

 

No, that is purely your opinion. It is far from fact. Back when I would even use ibt instead of linx I could get stable overclocks with ibt but then still fail with prime. As I said you can call it stable with linx all you like but if that is your only test then you've not tested it enough to truly call it stable.

 

As to your "experience" ... who cares? What matter is what you say and how you back it up. I've been overclocking for 15 years back when you had to manually unlock the Athlon Thunderbirds via a pencil ( or more complex/permanent ) trick. Does that instantly make everything I post correct? No it does not, just like your supposed experience doesn't make your OPINION on Linx correct.

 

You're the one who started this with a chip on his shoulder too. Over what? Whether or not Turbo boost counts as stock? Stating that it's easier to overclock than buy a new, faster cpu?

 

You constantly try to present your opinions as facts and now are trying to base it on what? 10 years of forum posting? GG.

 

I clearly pointed out the difference of ease in a previous post about overclocking vs buying but you've conveniently ignored all that and posted instead about your supposed experience and now moved on to making unfounded claims against stress testing.

 

And before you try turn my words against me I am for the record stating that I do not believe you can 100% call a high level overclock stable just running linx for 30 minutes and bypassing Prime95 entirely. I maintain you need to run prime95 to ensure 100% stability, I would say you should probably run both ( temperature stress testing on Linx is especially useful ) but I value Prime95 more for overall, complete stability. That is my opinion, you have yours, your opinion however != Fact.

 

Maybe in your reply you can also address the original point you made of how overclocking is supposedly easier than buying? Still waiting for your reply on that one before you went off on this tangent.

[MeNaCe-NZ]

Ah, good luck on that! I've had overclocked systems that survived days of Prime95, hours of OCCT/Linpack/LinX, but would crash 10-24 hours into their work in a render farm until they were clocked down some.

 

"True stable" is pretty hard to ascertain. If your machine crashes once in a year due to the OC, do you consider it stable? Or stable enough for you? And if "true stable" is the end goal, we should probably be using Xeons with some ECC and avoid the whole overclocking discussion to begin with! There's a reason Intel bins the way they do (well, a couple reasons, but we'll pretend they don't downclock higher-binned chips to meet market demand for the lower end CPU's).

 

True enough, stable is a matter of the eye of the beholder and it will come down to real world results but to me at least it's what the repercussions of any crashes etc. may be ( data loss, possible hardware fault etc. ). That's why I've always preferred Prime, it's more likely to find anything untoward than short IBT/Linx etc.

 

You are right though and none of that will go on to say you aren't going to get issues in rendering etc. and that it's truly 100% stable but I feel if that is the case one should probably do everything possible to determine stability before proceeding. Hence might as well run both, many people will use ibt/linx initially to get their first lot of high end settings then prime from there, get errors in prime then up voltage/downclock and start again.

[SuspectDown]

Maybe in your reply you can also address the original point you made of how overclocking is supposedly easier than buying? Still waiting for your reply on that one before you went off on this tangent.

 

I've already addressed that topic before, you must have skipped over those posts.

 

As for Linx WITH avx vs P95, you obviously have your outdated opinion and nothing is going to change that. Yes, P95 was the go to stress test back in the day... Not anymore. Using Intel's Linpack (OCCT, Linx, etc) and making sure AVX extensions are enabled is the best way to ensure stability nowadays. P95 neither stresses nor heats up the CPU as good as that since Sandy Bridge came out. I've come to to this conclusion from empirical evidence from overclocking Intel CPU's that are AVX compatible. P95 stable doesn't mean it's Linpack w/ AVX stable.

[MeNaCe-NZ]

I've already addressed that topic before, you must have skipped over those posts.

 

How is buying something from a shop easier than overclocking manually?

 

If you think you can judge your overclock as "stable" in 15-30 minutes then you don't know as much about over clocking as you think you do and it's no wonder you think it's "easy".

 

Not to mention it all requiring a person to have the correct hardware for over clocking to begin with. The majority of the time they are going to get better performance by upgrading their current hardware than trying to overclock existing hardware that will probably require a form of upgrade in the first place.

 

Cooling at least which requires getting into the case and removing the CPU ( clean off old TIM to apply new TIM, doing in the socket is just begging to damage something ) and a varying degree of difficulty depending on the cooler used ( which depends on how far you want/need to push your overclock ).

Add to this the amount of research on good/specific settings/firmware updates for their current motherboard ( if they didn't have to upgrade it ) and then voltage tweaks and the days of true stability testing and that's before you even look at if you need to start messing with your ram timings or not.

I'm sure I've missed even more than that too so to turn your question against you, how is buying from a shop harder than manually overclocking?

 

Your next reply:

 

Well, my experience ranges from air to phase change cooling (which I highly doubt more than 2-3 people on this forum if that have). I've hit significant overclocks with many platforms and am known on both EOCF and XS.

 

Yes, you can get a stable OC in 15-30 minutes. The days of having to run Prime95 for 8+ hours are over. 30 minutes of Linx w/ AVX instructions is all you need and I have built several high end systems overclocked with that regimen to prove it.

 

You addressed nothing, you went off on your own tangent.

 

As for Linx WITH avx vs P95, you obviously have your outdated opinion and nothing is going to change that. Yes, P95 was the go to stress test back in the day... Not anymore. Using Intel's Linpack (OCCT, Linx, etc) and making sure AVX extensions are enabled is the best way to ensure stability nowadays. P95 neither stresses nor heats up the CPU as good as that since Sandy Bridge came out. I've come to to this conclusion from empirical evidence from overclocking Intel CPU's that are AVX compatible. P95 stable doesn't mean it's Linpack w/ AVX stable.

 

Interestingly since my last post I thought to try hunt any information that would actually prove your point. I really can't find anything, almost everything is too the contrary in that you shouldn't solely rely on a Linx stress test to prove stability perhaps you can provide some information evidence that turns your opinion into fact?

 

Also Prime95 uses AVX ( for over a year ) or are you only picking and choosing when information is too old as such?

 

Linpack W/AVX doesn't mean it's Prime95 stable either ( Oh I better add in w/ avx since you seem to love writing this ).

[SuspectDown]

No, that is purely your opinion..

 

I figure you aren't going to shut up about linpack w/ AVX vs P95, so I did some small runs for you. Linpack w/ AVX vs. P95 w/ AVX Small FFT's vs. P95 w/ AVX Large FFT's. I ran P95 right after Linpack as well so there wasn't any bias and reset data between runs.

 

Yes, Linpack was a few degrees hotter. Yes, Linpack used a couple more watts. Most importantly, YES, Linpack stressed the system the most as evident by the largest VDROOP...

 

Linx: http://i.imgur.com/xiTamLW.png

P95 Small FFT: http://i.imgur.com/JVyGApC.png

P95 Large FFT: http://i.imgur.com/4DK7xMM.png

 

You can argue I manipulated the temperature of my WC loop (I didn't, all runs were at my 24/7 silent settings), you can argue I manipulated the ambient temperature (I didn't, all tests ran within +-1 degree Fahrenheit), you can argue a few watts extra doesn't matter (whatever), but what you CANNOT argue is the VDROOP difference.

[SuspectDown]

Interestingly since my last post I thought to try hunt any information that would actually prove your point. I really can't find anything, almost everything is too the contrary in that you shouldn't solely rely on a Linx stress test to prove stability perhaps you can provide some information evidence that turns your opinion into fact?

 

Also Prime95 uses AVX ( for over a year ) or are you only picking and choosing when information is too old as such?

 

Linpack W/AVX doesn't mean it's Prime95 stable either ( Oh I better add in w/ avx since you seem to love writing this ).

 

Oh, and I'll be happy to replicate the tests with longer stress times, different hardware, etc. You say the word

[MeNaCe-NZ]

I figure you aren't going to shut up about linpack w/ AVX vs P95, so I did some small runs for you. Linpack w/ AVX vs. P95 w/ AVX Small FFT's vs. P95 w/ AVX Large FFT's. I ran P95 right after Linpack as well so there wasn't any bias and reset data between runs.

 

Yes, Linpack was a few degrees hotter. Yes, Linpack used a couple more watts. Most importantly, YES, Linpack stressed the system the most as evident by the largest VDROOP...

 

Linx: http://i.imgur.com/xiTamLW.png

P95 Small FFT: http://i.imgur.com/JVyGApC.png

P95 Large FFT: http://i.imgur.com/4DK7xMM.png

 

You can argue I manipulated the temperature of my WC loop (I didn't, all runs were at my 24/7 silent settings), you can argue I manipulated the ambient temperature (I didn't, all tests ran within +-1 degree Fahrenheit), you can argue a few watts extra doesn't matter (whatever), but what you CANNOT argue is the VDROOP difference.

 

What exactly does that prove? How does any difference in VDroop to you prove that Linpack over 30 minutes is going to be better at proving stability than Prime95 over 8 hours? That was your claim after all and that basically noone need ever run Prime95 and 30 minutes of Linx would be fine.

 

Maybe you have some good in depth discussions on the topic ( stability testing of 30 mins Linx vs 8 hours Prime or similar timeframes ) I can read from the forums you a guru of? I'm happy to read over any past debates you must have had on the issue because you certainly seem sure of yourself so must have had many before to get to the bottom of the debate at hand.

 

That's the second time you've ignored the part of ease of over clocking vs buying so we'll just assume you gave that point up then.

[SuspectDown]

That's the second time you've ignored the part of ease of over clocking vs buying so we'll just assume you gave that point up then.

 

You just keep repeating the same question, and I already touched on it. Yes, you can make multiple posts saying the same thing, but I already touched on that topic.

[SuspectDown]

What exactly does that prove?

 

You said P95 is the go to stress test and I said linpack causes more stress. Yes, the time dedicated to stress testing is another discussion. What does it prove exactly you say? My screenshots prove that linpack w/ avx is causes the CPU to use more power, produce more heat, and stresses the system harder (as evidenced by the VDROOP). The screenshots prove that.

 

How can you say P95 is more stressful when it doesn't cause the same VDROOP? < Can you answer that? This whole thread has turned into off topic tangents because you keep attacking what I say, yet I provide legitimate proof and you just want to keep speaking rhetoric.

Edited August 7, 2015 by SuspectDown

[Eronenris]

The worst FPS drops I get are on Rishi, the lowest value being 23 FPS in the slums. Every planet that has expansive terrain, such as Balmorra or Alderaan, I also get large FPS drops. I play with every graphical variable maxed on a 2560 x 1440, 32'' monitor. My question is are the FPS drops the fault of the game or my computer? Every other game I play has a 50 - 60 FPS on maxed out settings (I use vertical Sync). If it's my computer, does anyone have a suggestion of how to increase the FPS in this game?

 

Here are my computer's specifications:

 

GPU: Geforce GTX Titan X

 

CPU: Intel® Core i7-2700k CPU @ 3.50 GHz

 

Memory: 16 GB Ram

 

And I use the Windows 7 Operating System.

 

It is definitely your computer, but not in the way that you think.

 

I only have an i5 3570k, 16 gigs of ram, and a gtx760. nothing top of the line by any means. Even on fleet I have zero lag. No server lag or FPS drops. But less than a week ago when i started playing this game again I had tons of lag everywhere I went. I did this one simple fix that took me about 30 minutes and ever since the game runs amazingly well. My entire computer runs better now.

 

So if you haven't tried this already i suggest that you do. Here is a link to my tomshardware post with the solution on it.

 

http://www.tomshardware.com/answers/id-2749756/lagging-games.html

 

The issue is caused by having many years worth of video drivers all on your machine at the same time and it causes conflicts in some software. This fix will revert your PC back to zero video drivers off any kind and then you just manually install the one most current driver. (don't use the geforce experience to get this driver)

Edited August 7, 2015 by Eronenris

[MeNaCe-NZ]

You just keep repeating the same question, and I already touched on it. Yes, you can make multiple posts saying the same thing, but I already touched on that topic.

 

Yes you touched on it, I countered you assertion, you failed to respond. As I said, we'll just assume I'm right and that buying an upgrade is much easier than overclocking. You clearly don't want to ACTUALLY PROVE your statement in any meaningful manner.

[MeNaCe-NZ]

You said P95 is the go to stress test and I said linpack causes more stress. Yes, the time dedicated to stress testing is another discussion. What does it prove exactly you say? My screenshots prove that linpack w/ avx is causes the CPU to use more power, produce more heat, and stresses the system harder (as evidenced by the VDROOP). The screenshots prove that.

 

How can you say P95 is more stressful when it doesn't cause the same VDROOP? < Can you answer that? This whole thread has turned into off topic tangents because you keep attacking what I say, yet I provide legitimate proof and you just want to keep speaking rhetoric.

 

OK see now you're just purposely being annoying it would seem.

 

Let's make this basic so you can stop going off on tangents.

 

I will do it with quotes and a basic summary, should be simple.

 

And yes, overclocking is easier than buying a new CPU... In the span of 15-30 minutes I can have a safe, reliable OC with a decent cooler from the comfort of my desk. How is getting a new CPU easier than that?

 

Your outlandish "easier" comments you've really failed to back up and you're rather ignorant claim that 15-30 minutes is all you really need to cool an OC stable.

 

If you think you can judge your overclock as "stable" in 15-30 minutes then you don't know as much about over clocking as you think you do and it's no wonder you think it's "easy".

 

My reply ( and the rest but that was more to the "ease" point ). Fairly self explanatory.

 

Yes, you can get a stable OC in 15-30 minutes. The days of having to run Prime95 for 8+ hours are over. 30 minutes of Linx w/ AVX instructions is all you need and I have built several high end systems overclocked with that regimen to prove it.

 

Your next claim to fame, again trying to imply a 15-30 min run of linx is all one needs to be called stable with their overclock.

 

From what I've known to have an overclock that is considered truly stable you really need to run the prime test . Whilst Linx is great for stress testing temperatures I wouldn't personally rely on it enough over 30 minutes to call my overclock completely stable.

 

Many people would run both though personally if I'm going to run a prime test anyway I am fairly confident it will find anything wrong that linx runs would have anyway. The same cannot be send about running linx over prime.

 

You could argue each to their own but if you are proclaiming the death of Prime testing or having it as a thing of the past you are just ignorant.

 

My reply - giving linx is kudos, as I said it's great for testing temps because it indeed does stress the cpu more intensly for it's short run, it is by no means to be considered a full test on stability.

A good 8+ ( 24 even better ) run on Prime95 can be considered far more reliable than a 15-30 minute run on Linx.

 

That is my stance, I think if you actually read my above post you will see I've even stated is opinion but you are basically trying to make your outlandish claim fact. You're the one who called Prime95 obsolete and no longer required yet have failed to prove it what so ever ( since most expert overclockers will run both as I also aluded to above ).

 

So being that you are the one making the claims ( as proven above ) how about you justify how you think a 15-30 min run of linpack/linx ANYTHING YOU LIKE is going to be better than a 8 HOUR run of Prime95 ... the extra stress above is hardly even much more than Prime95 to begin with, it's not even going to come close to recreating the possible stress a long render or hours gaming could generate. So instead of trying to turn this around on me, how about you try prove your own point - where is your evidence? I've even tried finding evidence for you on other forums but you're not even close to being in the realms of popular opinion let alone fact.

[ALaggyGrunt]

Prime95 is good because it checks your CPU's output while under stress against known results. You may not crash, but Prime95 will find out if your CPU is making random single-bit errors, which may or may not show up during normal usage, like when running Windows Update (or just installing Windows) and uncompressing stuff. Single-bit errors cause weird things to happen.

 

Oh, what was the offending program causing the FPS drops?

Edited August 8, 2015 by ALaggyGrunt

[BellsofGuilt]

I have good news concerning overclocking my CPU (an i7-5820k): I managed to overclock it to 4.0 GHZ and now the FPS problems I've been having with this game have vanished. Whereas before I was getting FPS as low as 22 in Rishi, now I get a constant 60 FPS (with slight variation from 57-62).

 

This vindicates everyone in this thread who said my CPU ws bottlenecking my GPU ( a Titan X) and it also might interest people who have the same problem I did.

Edited August 10, 2015 by BellsofGuilt

[Shar-Blade]

So I upgraded my PC to:

 

Geforce GTX980 Graphics card

Intel I7 4790K Chip

MSI Z97 Gaming 7 motherboard

16 Gigs of Ram

 

And the frame rate is just horrible. 30ish at best. Between 10 and 15 in flashpoints, ± 6 in 8 man ops. Around 1 in 16 man ops and 0.2 when trying the walker on Yavin...

 

Any advice why SWTOR would run so badly on a PC that should be able to run it without a sweat ?

 

 

I think so yes, the 4790k will overclock well also if you feel like looking into researching this and getting the relatively cheap extra cooling required. It still runs at an easy 4.0 Ghz on the stock cooling and that's nothing to be sneezed at and turbo's through to 4.4 on the same stock though temps can get up a bit so an aftermarker cooler wouldn't go amiss anyway really.

 

If you were highly worried about money I would recommend the i5 as it will run most games just as well as the i7 until hyper threading becomes more common place in game usage which I suspect we'll see more of with DirectX 12 in Windows 10.

 

The i7 is the more future proof of the cpu's, don't get the xeon, it would be crap for gaming comparatively and it's really not what they are designed for.

 

Of course also ensure the motherboard you are currently running supports the CPU you want to buy, I would probably recommend upgrading this at the same time ( if you didn't when you got the Titan which I doubt? ) to suit your current and future needs ( in terms of the bells and whistles different motherboards come with ).

With that CPU you'll be looking for any of the Z97 chipset boards, the difference in prices is often dictated by features ( on board audio, SLI - you may want to add another Titan later on lol ).

 

Some people might recommend waiting to upgrade to the new skylake cpu's coming out soon but honestly from the early leaked benchmarks I've read they won't really offer much of a gaming increase per se ( maybe a few FPS ) and possibly even more room for overclocking.

 

This was to be expected though really, CPU's are getting smaller and using less power but they aren't really pushing out more raw processing power to previous generations before them ( just check some of the halswell benchmarks vs previous generations equivalent CPU in gaming and the FPS difference is rather miniscule ideally ).).

[Rouakh]

Lol weird

 

i5-2500k

8 gb ram

760 gtx

 

45-55 fps on skyhigh game settings with 1920x1200 screen. Frame drop only due to lag spikes if it happens.

Edited September 3, 2015 by Rouakh