Haswell Pentium G3258 @ 3.8Ghz, better for gaming than a Q9400 @ 2.66 (quad)?

[Lepton87]

Is there any difference between the E3-1231v3 and the i7-4770 other than
(1) The i7-4770 has 100MHz higher turbo
(2) The E3-1231v3 has 4W lower TDP
(3) The E3-1231v3 has no IGP
(4) The E3-1231v3 can support ECC RAM
(5) The i7-4770 costs $60 more

Is there something I'm missing for why these two CPU's should be so different in price?

Maybe one thing, some cheap mobos might not support Xeons at all.

 

[Yuriman]

Some might find an IGP, ECC and 100mhz higher clocks to be worth $60. It's cheaper than most any current generation discrete video card.

 

[frozentundra123456]

i7 will be able to play games longer than an i5, even the mainstream version, I think that in the future even SB i7 could be better than HW i5 in games. There is one example of a game that wants more than 4 threads already, NBA 2K15. I upgraded my 2500K to 2600K just for this reason but the upgrade itch had to be scratched again so I have what's in my sig with my old system still waiting to be sold.

So are there any benchmarks for this game? I haven't been able to find anything except anecdotal comments that i5 is having problems. Even then, I am not sure if that is indicative of a trend or if this is just a very poorly coded game.

 

[SteveGrabowski]

So are there any benchmarks for this game? I haven't been able to find anything except anecdotal comments that i5 is having problems. Even then, I am not sure if that is indicative of a trend or if this is just a very poorly coded game.

I'd love to see that too, as NBA 2k15 is one of the games I'm most interested in. But it's kind of strange, as the recommended specs for the game are Core i5 or better.

http://store.steampowered.com/app/282350/

 

[cbn]

Here is a good example of what I mean, mostly 2 HW threads are enough but some games have started appearing that clearly want more threads.

It's not the norm but more and more games start to behave like that, necessitating at least a 4 threaded CPU

Since that test was done, I have to wonder if Nvidia made improvements through drivers. If so, by how much?

That was a relatively new game at the time and an outlier, like you mentioned, in the frame variance.

P.S. From my standpoint, it seems this stuttering issue is almost purely graphics in nature because as reported earlier lowering detail settings makes it go away. If it was cpu the problem would persist even at low resolution.

 

[cbn]

Makes me wonder if HT has proved so useful in Haswell i3's if it's not worth it to go for Xeon/i7 in case games want 8 threads in 2/3 years. Everyone says SMT is useless in games and to buy an i5, but SMT is obviously doing a lot of heavy lifting with the i3 Haswells.

Since most modern games are at least four threaded, having the HT on the dual core is almost always guaranteed to yield gains.

But on the quad core, I have read that having HT on when it is not needed can reduce performance. (So it really depends on how many threads these games have)

 

[Lepton87]

So are there any benchmarks for this game? I haven't been able to find anything except anecdotal comments that i5 is having problems. Even then, I am not sure if that is indicative of a trend or if this is just a very poorly coded game.

Yes, it is a purely coded game but the relationship between an i5 and an i7 is bound to look more and more like the relationship between Pentium and i3 looks now. It might take years but games are going to be better threaded.

 

[Yuriman]

Yes, it is a purely coded game but the relationship between an i5 and an i7 is bound to look more and more like the relationship between Pentium and i3 looks now. It might take years but games are going to be better threaded.

I suppose it'll be a race between the i5 and FX-8350. We'll see if single-threaded performance being so low that framerates are smoothly unplayable, or stuttering being so bad that regardless of how high framerates are, it's unplayable, will happen first.

 

[cbn]

I suppose it'll be a race between the i5 and FX-8350. We'll see if single-threaded performance being so low that framerates are smoothly unplayable, or stuttering being so bad that regardless of how high framerates are, it's unplayable, will happen first.

I would take the i5, because it appears the stuttering issue can be fixed by lowering graphics settings.

But a smoothly unplayable slow octocore does not have this option.

 

[lyssword]

So IPC advantage of so much more cache and better memory subsystem is on the order of 3% taking into account frequency. So you have 5% higher frequency and 2.8% higher score.

What's the memory speed you're running? I"m at 1333.

 

[AtenRa]

I would take the i5, because it appears the stuttering issue can be fixed by lowering graphics settings.

But a smoothly unplayable slow octocore does not have this option.

Lowering graphics will have an affect on both CPU.

 

[cbn]

I suppose it'll be a race between the i5 and FX-8350. We'll see if single-threaded performance being so low that framerates are smoothly unplayable, or stuttering being so bad that regardless of how high framerates are, it's unplayable, will happen first.

I would take the i5, because it appears the stuttering issue can be fixed by lowering graphics settings.

But a smoothly unplayable slow octocore does not have this option.

Lowering graphics will have an affect on both CPU.

Yes, Lowering graphics will increase FPS on both.

But I still think the cpu that produces high FPS (and micro stutters) is a better bet than a cpu that produces low FPS (but is smoothly unplayable) (re: At least the high FPS is there from the start with option #1. Who knows how much FPS could be raised with option #2?)

 

[sm625]

G3258 will wreck a Q9400 in almost every way. It's only weakness is multitasking under heavy load. The Q9400 is just so old, with its two separate cores talking over a FSB. Nehalem is a huge step up from Q9xxx, but I still ditched my nehalem i5 in favor of an overclocked G3258. It's so comically fast most of the time. It does stutter though. I've seen it, but its just so damn fast most of the time that I wouldnt go back.

 

[skipsneeky2]

I wouldn't upgrade from a Q9400 unless i was at least getting a i3 out of it. Having had a i3 2100 that chip certainly felt smoother then my old q6600@3Ghz ever did. No point in compromising in any department.

Guessing any recent i3 would destroy the i3 2100.

 

[crashtech]

I wouldn't upgrade from a Q9400 unless i was at least getting a i3 out of it. Having had a i3 2100 that chip certainly felt smoother then my old q6600@3Ghz ever did. No point in compromising in any department.

Guessing any recent i3 would destroy the i3 2100.

Yeah. improvements in HT plus a speed bump from 3.1 to 3.4GHz means the 4130 is around 30% faster than the 2100 fully loaded, generally speaking.

 

[skipsneeky2]

Yeah. improvements in HT plus a speed bump from 3.1 to 3.4GHz means the 4130 is around 30% faster than the 2100 fully loaded, generally speaking.

Newegg is silly with their i3 pricing, there is a 4150 which is 3.5GHz and it's priced the same as the 4130.There is a $160 4370 which clocks at 3.8GHz.Guess older games like Counter Strike would fly on that.Sucks $30 more gets you a lowly clocked 3.2GHz 4460 i5.

 

[Cerb]

Newegg is silly with their i3 pricing, there is a 4150 which is 3.5GHz and it's priced the same as the 4130.There is a $160 4370 which clocks at 3.8GHz.Guess older games like Counter Strike would fly on that.Sucks $30 more gets you a lowly clocked 3.2GHz 4460 i5.

That's not Newegg. That's Intel. The i3-4130 is old, and the i3-4150 has replaced it, just as that 3.2GHz i5 replaced a 3.0GHz one.

For most uses, the lower-clocked i5 will be a better choice, though the 4590 is clearly the go-to non-OC model.

 

[escrow4]

That's not Newegg. That's Intel. The i3-4130 is old, and the i3-4150 has replaced it, just as that 3.2GHz i5 replaced a 3.0GHz one.

For most uses, the lower-clocked i5 will be a better choice, though the 4590 is clearly the go-to non-OC model.

I'd say 4690 - it will boost up to 3.9GHz, the 4590 tops out at 3.7GHz. Base is 200MHz higher too.

 

[Cerb]

I'd say 4690 - it will boost up to 3.9GHz, the 4590 tops out at 3.7GHz. Base is 200MHz higher too.

i5-4460->4590: $10
i5-4590->4690: $25

With AMD's offerings being meh, they all work out pretty well.

 

[jacktesterson]

I'd say 4690 - it will boost up to 3.9GHz, the 4590 tops out at 3.7GHz. Base is 200MHz higher too.

200 MHz is not worth the price increase IMO.

I agree the 4590 is the best bang for the buck CPU from Intel for gaming. (If O/C isn't involved)

 

[crashtech]

Higher end CPUs hold their value better. For those who will resell, this can reduce or erase the price differential over time.

 

[escrow4]

200 MHz is not worth the price increase IMO.

I agree the 4590 is the best bang for the buck CPU from Intel for gaming. (If O/C isn't involved)

I'd take the extra $25 as you won't be overclocking so you may as well start off with the highest boosting chip.

 

[cbn]

Games don't use AVX instructions

Apparently GRID 2 used in the Tom's hardware Pentium G3258 review does use AVX, but in general, you are right in that it appears games using AVX are not common.

 

[cbn]

Looking at the Tom's Hardware Pentium G3258 review, here are results from Metro Last Light (which was the oldest game to scale quad core used in the test):

http://www.tomshardware.com/reviews/pentium-g3258-overclocking-performance,3849-7.html

Looking at the above results, the frame time variance appears to scale inversely with frame rate with HW thread count differences not being a noticeable factor. This, in contrast, to some of the other newer games in Tom's test. Therefore I'm thinking some of the frame time variance issues are related to graphics drivers.

P.S. Only Tomb Raider and World of Warcraft were older, but they appear to not scale with cores. (But look how low the frame time variance is (in general) and how close the frame variance are for the OC G3258 and some of its competitors)

http://www.tomshardware.com/reviews/pentium-g3258-overclocking-performance,3849-9.html

http://www.tomshardware.com/reviews/pentium-g3258-overclocking-performance,3849-10.html

 

[Yuriman]

Much appreciated, cbn.