GeForce Titan: 2500K vs 4790K (66 games tested)

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

NTMBK

Lifer
Nov 14, 2011
10,269
5,134
136
Instead of pissing around 'refining' their fake cores, why don't they just stop being such tight-arses and give the mainstream more real cores. Same fun without the trade-off, happy days for all.

Why not both? *shrug* In order to squeeze even more IPC out of their core, Intel needs to keep making it wider- which means that there is a lot of CPU core going unused most of the time. Hyperthreading is a great way to squeeze more performance out of what is already there.
 

sxr7171

Diamond Member
Jun 21, 2002
5,079
40
91
So I'm good for another 2 years of 1080p gaming. Cool.


I have the same CPU at 4.8Ghz and same mobo and my reading of this thread is making me think the opposite. I can't deny I have those frame rate drops in Crysis 3 that are being spoken of. Also I wonder if my Watch Dogs stuttering issues may be alleviated by having HT. This is the first time since I bought the CPU that I am doubting my choice of i5 over i7.
 

BrightCandle

Diamond Member
Mar 15, 2007
4,762
0
76
I have the same CPU at 4.8Ghz and same mobo and my reading of this thread is making me think the opposite. I can't deny I have those frame rate drops in Crysis 3 that are being spoken of. Also I wonder if my Watch Dogs stuttering issues may be alleviated by having HT. This is the first time since I bought the CPU that I am doubting my choice of i5 over i7.

Watchdogs long stutters are caused by storage reads. The frame rate stutters are caused by GPU VRAM limits. Having 6 real cores and 6 virtual ones does not do anything to alleviate the stutters for me.
 

BD2003

Lifer
Oct 9, 1999
16,815
1
76
To say that older CPUs have no issue maintaining 60fps would be a grave misconception. And 120fps? Targeting a sustained 120fps is a fool's errand, take it from one such qualified fool. It's almost absurd to consider X CPU vs Y CPU for a 120fps target, because all you're looking at is varying degrees of miserable failure. There is a tiny minority of games in which 120fps can be realistically maintained, and then subtract from that the handful of games with engines that malfunction at frame rates that high - you're not left with much. The stagnant CPU industry hasn't helped.

That's not true at all. You just have to accept a lower resolution or level of effects, or buy a hell of a lot of graphics cards. The vast majority of games I've played since I got my 120fps monitor handle it just fine.

The problem is that you cant just buy more CPUs, and most settings are more involved with the GPU than the CPU. Every frame depends on the CPU being able to feed the GPU, and a lot of times you hit a wall on single threaded performance when you're pushing that many FPS. And 120fps is only a niche for now, most monitors are going that way (and higher) eventually. VR is going to demand more like 150-170fps. (75/85hz per eye).

The reason we see such a tight grouping between all the CPUs, is because the tests are being done with settings targeted at 60fps. Those low settings tests arent irrelevant, they're real world settings for people who want to push framerate instead of resolution.
 

toyota

Lifer
Apr 15, 2001
12,957
1
0
plenty of games cannot even average 120 fps much less stay above it no matter how much gpu power that you have. heck some games can barely stay above 60 fps with the fastest cpus out there.
 

PPB

Golden Member
Jul 5, 2013
1,118
168
106
Why not both? *shrug* In order to squeeze even more IPC out of their core, Intel needs to keep making it wider- which means that there is a lot of CPU core going unused most of the time. Hyperthreading is a great way to squeeze more performance out of what is already there.

That is a bad logic being used, as Intel not only is making their cores wider, but it is also improving other areas for better efficiency and branch prediction. Haswell is proof that even by making the core wider, HT performance decreased and we can no longer talk about a 30% boost from HT, more like 20% these days.
 

Jovec

Senior member
Feb 24, 2008
579
2
81
30% boost from HT, more like 20% these days.

Has almost nothing to do with cores and HT and almost everything to do with optimization and the nature of the problem being solved in code.
 

2is

Diamond Member
Apr 8, 2012
4,281
131
106
I'd be tempted to add another Titan with many of those results.

That said, i'd like to see the minimums and not so much the difference in average fps.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,980
126
And I'll qualify my last post by saying that I don't accept 'lowering levels of effects' as a valid strategy for maintaining 120fps. I was taking it for granted that people don't want their games looking like arse as long as it means high frame rates.
Do you run 4K + 8xSSAA in every game? Because if not, you too are lowering details to get a higher framerate.

Crysis 1 looks like utter ass without AA.

I think it's silly to automatically turn all settings up to 11 and then complain about low performance. Tweaking is one of things that makes up PC gaming. If you're not going to adjust settings, you may as well just buy a console.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,980
126
what does AA have to do with what I was saying? I said it drops below 60 fps with a 2500k and that is FACT. so your logic as usual is that I should turn up the AA until I am gpu limited instead of cpu limited? who cares in this particular case as the point is that a stock 2500k itself does not have the ability to keep the framerate from low 50s or even high 40s. most people buy a high end gpu to stay above 60 fps and you cant do that in the old Crysis which again was the point here.
I don't think you understand what I'm saying. What I'm saying is that at my settings I'm sure I'd be getting more performance than you if I took a screenshot where you did.

I can't currently locate that spot, otherwise I'd have already posted something. Hence the reason I posted the other one.
 

PPB

Golden Member
Jul 5, 2013
1,118
168
106
Has almost nothing to do with cores and HT and almost everything to do with optimization and the nature of the problem being solved in code.

Until you find this behaviour using the same piece of software on both CPUs.
 

BD231

Lifer
Feb 26, 2001
10,568
138
106
BFG can you run some game streaming tests with OBS? Preview feature would be enough. Preferably 1080p down scaled to 480p or lower @ 30fps on x264 very high pre-set?

Would be nice to see if the IPC gains did anything for something that's actually CPU intensive.
 

BFG10K

Lifer
Aug 14, 2000
22,709
2,980
126
Oh come on, there's a difference between degrading draw distances, physics and other effects, and merely not running the game at unreasonably high resolutions with crippling levels of AA.
Really? Unreasonable? Crippling? Are those facts or your opinion?

Why can you reduce AA and resolution and claim you're not degrading IQ, but I can't reduce objects and claim the same? Who made that rule and when?

I wouldn't argue with that, but what's your point?
What AA are you using in the game?

I could just as easily say "turning all settings up to 11 is one of the things that makes up PC gaming.
You're running settings outside of your spec and complaining that things are too slow. That's not PC gaming, that's foolish.

If you're not going play at max detail, you may as well just buy a console", we'd both be engaging in hyperbole. And while I agree that tweaking is part of the PC experience, I prefer to tweak to the enhancement of IQ, not to it's detriment.
By "max detail" I take it you mean "only the settings I care about". That's actually not max detail.

As for reducing CPU intensive settings, as even BD2003 admitted, these are generally scarce (I was trying to play BF3 on a q9650 with my GTX 780 a while ago and there was less than 5fps difference between all settings max vs min because they simply had ah heck all effect on CPU load), and when you do find them, they tend to have a disproportionately dramatic impact on IQ (for example, 'Objects' in Crysis is one of the CPU dependant settings, and draw distance for objects is one realm where Crysis is really starting to show it's age and can ill afford to have it reduced).
You have that backwards - very high objects have a disproportionately dramatic impact on performance while the difference in IQ is negligible, unlike SSAA which reduces the shimmerfest the game produces over the entire scene, hence providing a massive improvement in IQ.

I'll gladly give up 5 distant bushes/rocks in exchange for not having to look at swarming bees on the screen.
 

BD2003

Lifer
Oct 9, 1999
16,815
1
76
You said it.



And I'll qualify my last post by saying that I don't accept 'lowering levels of effects' as a valid strategy for maintaining 120fps. I was taking it for granted that people don't want their games looking like arse as long as it means high frame rates.


I'm at the point where 60fps looks like "arse" to me, so I'm willing to sacrifice what I have to to maintain 120. Not always possible as we both agree, but that's where your CPU comes in. I'm just a little frustrated at how quickly low resolution testing methodologies are brushed off at as being irrelevant to real world scenarios, and the immediate conclusion is drawn that CPU performance barely matters. IMO 120hz is just as relevant if not more so than 1440p or 2160p.

There's a massive gulf in requirements jumping from 60-120fps....and current CPU performance doesn't always get us there in the newest, most taxing games. Every monthly buyers guide on AT, Tom's, wherever, speaks as if there's almost no point to a faster CPU because it's only a few % points different, completely ignoring how much that picture would change with an alternative set of real world considerations (and choosing settings to match.)
 

Carfax83

Diamond Member
Nov 1, 2010
6,841
1,536
136
I have the same CPU at 4.8Ghz and same mobo and my reading of this thread is making me think the opposite. I can't deny I have those frame rate drops in Crysis 3 that are being spoken of. Also I wonder if my Watch Dogs stuttering issues may be alleviated by having HT. This is the first time since I bought the CPU that I am doubting my choice of i5 over i7.

HT will improve performance in the game yes, but like BrightCandle said, the stuttering is caused by the game's poor memory handling so that has nothing to do with the CPU.

The game doesn't use enough system ram, and instead goes to storage too much for data..
 

BD2003

Lifer
Oct 9, 1999
16,815
1
76
The more I think about it, the more I'm inclined to suspect that it is indeed a fact that 4K + full scene 8xSSAA is impossible on any currently available consumer GPU solution at playable framerates.



The problem is that you're trying to use this completely open-ended criteria. There's practically no limit to the resolution one can use, or the amount of AA one can pile on to a game. The equivalent would be if I expanded my criteria to include all potential .ini modifications etc. You can't tell me, with a straight face, that playing a game with 4K + 8xSSAA is as realistic a proposition as playing a game with all it's in-game detail* settings configured to their maximum values



I think I had settled on 4xSGSSAA. I spend hours testing various AA schemes before I gave up on my first Haswell system (courtesy of Asus' z87 RTC bug), and now I'll have to do it all over again because I neglected to document the outcome.

I don't suppose you've had any success with SGSSAA in DX10? As I remember it, I was having difficulty with blur, and looking through my my screenshots, all the ones with SGSSAA were taken in DX9.




*When I say 'details', I mean standard finite menu options like draw distances, physics, etc, excluding the potentially infinite realm of resolution and filtering. Sorry if this wasn't clear. I wouldn't call something like AA 'detail', as it doesn't add any new detail, but enhances existing detail (not to downplay it's importance). I guess one could argue semantics (eg. higher resolution can resolve finer, and arguably 'more', detail purely by virtue of having more lines) , but I've now outlined my use of the terminology.



I agree that SSAA represents a massive improvement to IQ in transparency-heavy games like Crysis. It's difficult to really appreciate the value of it until you've been playing with it on for a few hours and switch back to the crawling mess that was before.

But I don't agree on the objects being negligible, and I don't know why you choose to compare that setting to SSAA, which is entirely GPU dependant. My last post was clearly relating to CPU dependant settings. You don't 'give up' CPU intensive detail for GPU intensive filtering.



I can definitely sympathise with that. My problem is that I want it both ways, and I guess that dooms me to perpetual discontent.

I also agree with you on the low res testing, but for slightly different reasons.

Yeah, I cant stand the crawlies, TXAA pretty much solves that problem with relatively little performance impact compared to SSAA in the few games that support it. Too bad more don't (thanks NVIDIA!).
 

Arkadrel

Diamond Member
Oct 19, 2010
3,681
2
0
If you enjoy emulation then the haswell cpu upgrade is well worth it.


Basically for emulation, cpu-performance wise (dolphin/pcsx2/mame ect)

3.4ghz haswell = 4.4 ghz Ivy bridge = 4.6 ghz sandy bridge.
3.5ghz haswell = 4.6 ghz Ivy bridge = 4.8 ghz sandy bridge.


I think its funny a small i7-4700MQ (mobile quad cpu, 47w) with a tiny OC (+100mhz) is faster than a i7-2500k@4,8ghz (big desktop cpu) (1,1ghz oc or so), when it comes to this task.


So atleast for emulation the update from sandy bridge to haswell has been noticeable, around ~30-35% pr mhz.
 

sxr7171

Diamond Member
Jun 21, 2002
5,079
40
91
Watchdogs long stutters are caused by storage reads. The frame rate stutters are caused by GPU VRAM limits. Having 6 real cores and 6 virtual ones does not do anything to alleviate the stutters for me.

I see that's good to know. I have the game on on 840evo SSD. And also have 16GB RAM why won't it use the RAM better? I remember reading someone who put the game on a RAMdisk had less stuttering. As for VRAM, I'm hearing even users with 6GB VRAM have stutters. I'll do just about anything to get rid of them.
 

sxr7171

Diamond Member
Jun 21, 2002
5,079
40
91
HT will improve performance in the game yes, but like BrightCandle said, the stuttering is caused by the game's poor memory handling so that has nothing to do with the CPU.

The game doesn't use enough system ram, and instead goes to storage too much for data..

Could that be an "easy" fix for the developers? I heard that someone who put the game on a RAMdisk had less stutters.
 

sxr7171

Diamond Member
Jun 21, 2002
5,079
40
91
Yeah, I cant stand the crawlies, TXAA pretty much solves that problem with relatively little performance impact compared to SSAA in the few games that support it. Too bad more don't (thanks NVIDIA!).

Me too. Can't stand them and I use TXAA on all games that support it.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |