Support NeoGAF

[SleazyC]

Okay so as a 2500k owner, it's probably time? I'm OC to 4.6 I think but upgrading sounds like it's worth it, finally

Got a 980ti so I'm set on video card at least but gotta upgrade everything else :/

What resolution are you playing at?

I think you could probably survive on a 2500k/2600k but I think you'd see tangible benefits from moving to Skylake just from a platform perspective as well. You get access to USB3.1 and faster SSDs as well as DDR4 and its higher memory bandwidth.

 

[TouchMyBox]

Ay this is hilarius

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/10



A regression in performance over a 2 year old proc lmao

I don't see what this has to do with Pynchon.

 

[Zaptruder]

$524 5820K @ Budget PC Vic (or $559 at pccasegear)

+

$399 MSI X99A SLI Krait Edition (comes with USB 3.1)

or if you don't care about USB 3.1 you can get a cheaper MB from PCCaseGear.

Yeah, I'm done waiting. Picking up a cheaper mobo just because it's available in WA.

 

[SneakyStephan]

I actually had a similar thought. My i5-2500k still performs well but I'm afraid it'll shit the bed sooner or later as I've had it for over 4 years. Maybe trying to sell it while it's still operational would be the smarter choice.

Cpus don't break unless you've ran them at a much higher than stock voltage for a long time.
They last for absolutely ages

 

[Kiyo]

Do those 25% actually translate to in-game performance though or is that just benchmark wank?

"In CPU based benchmarks"

So no. If you look at the gaming benchmarks you see that it performs about equal to a 4790k and worse in some cases.

 

[Thrakier]

From Anandtech's review:



You're looking at a 25%+ increase going from Sandy Bridge to Skylake it seems. Still, it doesn't seem like there's much here for people who are on Ivy Bridge or newer. I'm really disappointed, I thought there would have been a bigger leap.

25% more isn't that much if you don't even need 1% right now.

 

[Zombie James]

Depends on your GPU and resolution.

https://www.youtube.com/watch?v=38KSD9o-LOs

Well there we go, a straight 2600K vs 6700K gaming comparison.

 

[marjo]

I've got a 2500k, and I'm definitely not going to upgrade. A 25% performance increase in synthetic benchmarks is extremely disappointing.

Real gaming benchmarks are even worse, showing very little difference, even with a high end graphics card at lowish (1080p) resolutions!

 

[Grief.exe]

How are the M.2 SSDs these days?

 

[tycoonheart]

Depends on your GPU and resolution.

https://www.youtube.com/watch?v=38KSD9o-LOs

So ya, pointless to upgrade especially if you game at 1440p.

 

[Mr Swine]

Jeez, I was afraid that these new CPU's from Intel wouldn't get that much performance boost over the last generation but none and sometimes worse? How is that even possible? What is Intel doing with their CPU's?

 

[Kiyo]

Jeez, I was afraid that these new CPU's from Intel wouldn't get that much performance boost over the last generation but none and sometimes worse? How is that even possible? What is Intel doing with their CPU's?

Giving AMD room to hopefully catch up and give them some competition. Pretty disappointed with these reviews/benchmarks as well. But it does make me feel a lot better about getting my 4790k last year instead of waiting.

 

[SolidSnakeUS]

I am currently on a quad core i5-3750K but at stock. Should I get a new cooler and OC it, or do you think I should get one of these so I'm back with an i7 machine? Basically, is it worth upgrading just so that I have an i7?

 

[Zombie James]

Jeez, I was afraid that these new CPU's from Intel wouldn't get that much performance boost over the last generation but none and sometimes worse? How is that even possible? What is Intel doing with their CPU's?

The door's wide open for AMD to make a splash with Zen next year. Intel's been stagnant for far too long.

 

[jabuseika]

No reason to upgrade from a 4690k.

 

[digitalrelic]

Feeling great about my 4790k @4.8ghz right about now.

Looks like I'll be safe waiting a few years to upgrade this beast.

 

[LilJoka]

If it isn't clear enough already, it's not wise to upgrade from sandy bridge (2500/2600k).

DDR4 has no benefit.
USB speeds is about it, if 80MB/s isn't enough on USB 3.0 then you are doing something special.

The door's wide open for AMD to make a splash with Zen next year. Intel's been stagnant for far too long.

Intel has consistently reduced its core architecture power consumption whilst boosting its IGP performance and maintaining CPU performance. If you really think Intel has been standing still, you are missing the bigger picture. Intel are trying to create faster CPUs at lower TDP all while trying to beat AMDs APU GPU performance.

 

[shiftplusone]

What resolution are you playing at?

I think you could probably survive on a 2500k/2600k but I think you'd see tangible benefits from moving to Skylake just from a platform perspective as well. You get access to USB3.1 and faster SSDs as well as DDR4 and its higher memory bandwidth.

Currently 1440p and expecting to continue with that with a gsync monitor this winter

 

[Mr Swine]

Giving AMD room to hopefully catch up and give them some competition. Pretty disappointed with these reviews/benchmarks as well. But it does make me feel a lot better about getting my 4790k last year instead of waiting.

I don't think they are giving AMD some room but probably have hit a wall what they can do with their architecture. Hopefully they can start from zero and make a brand new CPU on a new architecture.

I'm still running a i5-3570k and I'm wondering if it's worth upgrading to a 3770k

 

[SleazyC]

How are the M.2 SSDs these days?

Some of them are absurdly fast....

Boot times into Windows, synthetics, copying files and working with IO heavy applications seem to show a huge edge but not 100% sure if that translates to a anything noticeable in games.

 

[Kiyo]

I am currently on a quad core i5-3750K but at stock. Should I get a new cooler and OC it, or do you think I should get one of these so I'm back with an i7 machine? Basically, is it worth upgrading just so that I have an i7?

It depends what you do? If you're just talking about gaming then no, it's probably not worth the upgrade.

 

[GeoNeo]

Jeez, I was afraid that these new CPU's from Intel wouldn't get that much performance boost over the last generation but none and sometimes worse? How is that even possible? What is Intel doing with their CPU's?

You are starting to see the problems of Intel hitting instructions per cycle brickwalls.

That is why in their other platform (x99) and even Xeon platforms they are opting to go for more cores.

I am currently on a quad core i5-3750K but at stock. Should I get a new cooler and OC it, or do you think I should get one of these so I'm back with an i7 machine? Basically, is it worth upgrading just so that I have an i7?

Get a nice cooler and OC that CPU.

 

[drotahorror]

Maybe AMD's Zen will be a great upgrade next year. I can't believe it's been nearly 4+ years since Sandy Bridge and there still isn't anything out worth upgrading to (for gaming) really.

 

[SleazyC]

Currently 1440p and expecting to continue with that with a gsync monitor this winter

You're probably ok staying on Sandy Bridge; but I'd say you also have enough non-CPU benefits to upgrade if you wanted too.

I'd hold off to see what kind of performance users are getting when they have the CPUs in their hands. Maybe you'll see benefit from the games or applications you might be using.

 

[SneakyStephan]

Depends on your GPU and resolution.

https://www.youtube.com/watch?v=38KSD9o-LOs

This video actually shows it would be a worthy upgrade (though the performance increase is still incredibly dissapointing, just like it was with haswell and ivy)

The frametimes are much more even in the skylake graphs, which means way less stutter.

Coming from a phenom II (and now using a 4690k), inconsistent frametimes are literally the worst thing for gaming.
The stutter on my phenom II was unbearable even in games where I had good average fps.
imo any significant improvement in frametime consistency is worth an upgrade.

 

[LilJoka]

Maybe AMD's Zen will be a great upgrade next year. I can't believe it's been nearly 4+ years since Sandy Bridge and there still isn't anything out worth upgrading to (for gaming) really.

There is no bottleneck, 95% of games are GPU limited, so what would be the point in upgrading to something faster anyway.

 

[Zombie James]

This video actually shows it would be a worthy upgrade (though the performance increase is still incredibly dissapointing, just like it was with haswell and ivy)

The frametimes are much more even in the skylake graphs, which means way less stutter.

Coming from a phenom II (and now using a 4690k), inconsistent frametimes are literally the worst thing for gaming.
The stutter on my phenom II was unbearable even in games where I had good average fps.
imo any significant improvement in frametime consistency is worth an upgrade.

Frame time is a way better metric for performance than FPS alone. I wonder why Anandtech hasn't followed Tech Report's lead on this.

 

[drotahorror]

There is no bottleneck, 95% of games are GPU limited, so what would be the point in upgrading to something faster anyway.

True good point.

 

[Grief.exe]

Some of them are absurdly fast....

Boot times into Windows, synthetics, copying files and working with IO heavy applications seem to show a huge edge but not 100% sure if that translates to a anything noticeable in games.

I mainly need to get out of my 128 GB SSD to 256 or 512, but this is some encouraging performance data.

Our third real-life test covers disk activity in a gaming environment. Unlike the HTPC or Productivity trace, this one relies heavily on the read performance of a drive. To give a simple breakdown of read/write percentages, the HTPC test is 64% write, 36% read, the Productivity test is 59% write and 41% read, while the gaming trace is 6% write and 94% read. The test consists of a Windows 7 Ultimate 64-bit system pre-configured with Steam, with Grand Theft Auto 4, Left 4 Dead 2, and Mass Effect 2 already downloaded and installed. The trace captures the heavy read activity of each game loading from the start, as well as textures as the game progresses. In this trace we recorded 426MB being written to the drive and 7,235MB being read.

In our read-intensive Gaming trace, the Samsung SM951 posted an impressive 1,458.62MB/s, 29,995.25 IOPS, and an average latency of only 0.225ms

 

[SneakyStephan]

Frame time is a way better metric for performance than FPS alone. I wonder why Anandtech hasn't followed Tech Report's lead on this.

Because anandtech is a C tier site like tom's hardware

If you want decent hardware impressions and benchmarks you need to read pc perspective

 

[Evo X]

Mainly, I want to upgrade my 2500k not just for raw performance, but also the new tech that the latest mobos have over my Z68. Very interested in M2 SSDs, more bandwidth for Titan X SLI, and better frame times in games. Also, my Creative X-Fi Titanium doesn't work in Windows 10, so need a better on board audio solution.

From the guru3d review:

Combined with the series 100 chipset new features are available as well, SATA3, M.2, some manufacturers will even provide U2 and Sata Express. Then there's the added benefit of DDR4 memory that not only uses less power but the frequency can be so much higher as well, bringing more bandwidth and overall performance to the applications that require fast memory. Combine that with things like nice Gigabit jacks, exemplary audio solutions on the new motherboards and things like USB 3.1. So what i am trying to say here is that the overall platform experience is what it is all about for Skylake. Performance with kick-ass features.

 

[GeoNeo]

Frame time is a way better metric for performance than FPS alone. I wonder why Anandtech hasn't followed Tech Report's lead on this.

I'd also like to see the frame time comparisons take into account overclocking of these older chips to see how the stack up and even overcome any issues while running at stock.

End of the day people should upgrade not only for raw performance but the better chipset features etc.

 

[SolidSnakeUS]

Maybe AMD's Zen will be a great upgrade next year. I can't believe it's been nearly 4+ years since Sandy Bridge and there still isn't anything out worth upgrading to (for gaming) really.

I really hope it does come out swinging. AMD NEEDS to get back into the desktop CPU scene swinging and fast.

 

[SleazyC]

I mainly need to get out of my 128 GB SSD to 256 or 512, but this is some encouraging performance data.

Can always RAID0 together smaller SSDs and get even more stupid speed out of these M.2 SSDs. I think I was looking up benchmarks of RAID0'd SM951's and they were doing crazy stuff like breaking some benchmarking software.

 

[Rootbeer]

Mainly, I want to upgrade my 2500k not just for raw performance, but also the new tech that the latest mobos have over my Z68. Very interested in M2 SSDs, more bandwidth for Titan X SLI, and better frame times in games. Also, my Creative X-Fi Titanium doesn't work in Windows 10, so need a better on board audio solution.

From the guru3d review:

I'm in the same boat. rather skylake is more than only 5 or 10% faster than the previous year's offerings isn't the full story for me. I'm on an i7 920-based system and I want to move to a new system that not only has the latest processor, but all the other bells and whistles i'm going to need in a system in the coming few years. All of which is mentioned in your quoted text.

And when moving from an 920 to a 6700K, all those small gains from the last few years add up to be a REALLY compelling upgrade. I'm excited.

 

[Grief.exe]

I wonder if the i7 processors will really shine under DX12.

 

[cebri.one]

Broadwell IGP crushing Skylake's

 

[Coulomb_Barrier]

Jesus, this is even worse than I imagined.

Broadwell to Skylake (DDR4): Average ~2.7% Up

Oh dear. Typically with an architecture update we see a bigger increase in performance than 2.7% IPC. Looking at matters purely from this perspective, Skylake does not come out well. These results suggest that Skylake is merely another minor upgrade in the performance metrics, and that a clock for clock result compared to Broadwell is not favorable. However, consider that very few people actually invested in Broadwell.

http://anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/9

What's worse is in their other comparison between Haswell to Skylake, the 5% gain is only over the 4770k, not 4970K (Devil's Canyon).

 

[Ty4on]

Dat 5775C, tho.

 

[SneakyStephan]

Dat 5775C, tho.

Impressive

More interested in the FX 8350
This is the kind of thing I saw on my phenom II as well, crappy frametime consistency

 

[Ty4on]

Impressive

More interested in the FX 8350
This is the kind of thing I saw on my phenom II as well, crappy frametime consistency

Cache seems to have been Bulldozer's Achilles' heel as well.

Single threaded perf in cinebench was also 80% higher in Skylake than the 8350. Makes Zen's 40% sound less impressive and they never clarified that it was single threaded, only per core (Zen has multi threading).

 

[Tovarisc]

Speed of memory affects Skylakes gaming performance.

 

[Kiyo]

Speed of memory affects Skylakes gaming performance.

*snip*

Only by that much at those absurdly low resolutions that no one interested in gaming is actually playing at. Are there benchmarks that show the ram performance difference at 1080p/1440p with a discrete GPU?

 

[digitalrelic]

Speed of memory affects Skylakes gaming performance.

Why are these compared @ 640x480 on low settings? Since when is that relevant in the real world?

 

[Tovarisc]

Only by that much at those absurdly low resolutions that no one interested in gaming is actually playing at. Are there benchmarks that show the ram performance difference at 1080p/1440p with a discrete GPU?

Why are these compared @ 640x480 on low settings? Since when is that relevant in the real world?

Resolutions are low because they want to kill any GPU bottleneck there could be and get pure CPU+RAM performance tested. Sure that difference doesn't translate 1:1 into e.g. 1440p gaming, but clearly indicates that Skylake performance scales upwards as speed of memory increases.

 

[SleazyC]

Only by that much at those absurdly low resolutions that no one interested in gaming is actually playing at. Are there benchmarks that show the ram performance difference at 1080p/1440p with a discrete GPU?

Why are these compared @ 640x480 on low settings? Since when is that relevant in the real world?

You try to remove the impact of the discrete GPU when testing the performance the CPU has on gaming.

Of course in a real-world situation the GPU exerts an inordinate amount of impact on performance and its why you see benchmarks using a 980Ti and a Skylake CPU not showing much impact on gaming benchmarks.

EDIT: Beaten

 

[Evo X]

Only by that much at those absurdly low resolutions that no one interested in gaming is actually playing at. Are there benchmarks that show the ram performance difference at 1080p/1440p with a discrete GPU?

Why are these compared @ 640x480 on low settings? Since when is that relevant in the real world?

The purpose of testing at low res is to remove any GPU bottleneck so the differences in CPU architecture become more apparent.

And although faster RAM may not increase avg FPS much, it will most certainly lead to faster loading times, especially when paired with an SSD.




This is what bit-tech has to say on the matter. Upgrading largely depends on your budget and use for the system.

Looking past the last three generations (four if you include Broadwell which is sadly absent from the graphs as we're yet to obtain a sample), the answer to the question of whether you should upgrade is clear - yes. Sandy Bridge has been outdone in every test and in many cases by significant margins and even hefty overclocks can't help in some situations. However, in games, the difference isn't huge - it's photo/video editing and rendering where the real performance increases creep in.

 

[Tovarisc]

You try to remove the impact of the discrete GPU when testing the performance the CPU has on gaming.

Of course in a real-world situation the GPU exerts an inordinate amount of impact on performance and its why you see benchmarks using a 980Ti and a Skylake CPU not showing much impact on gaming benchmarks.

EDIT: Beaten

I would like to see Skylake dGPU tests in 1440p with hi-end GPU and different speeds of DDR4. There is now some hints that speed of DDR4 affects Skylakes gaming performance so now we need benches in "real life resolutions". For e.g. Anand only ran their dGPU tests with slowest DDR4, 2133MHz.

 

[SleazyC]

I would like to see Skylake dGPU tests in 1440p with hi-end GPU and different speeds of DDR4. There is now some hints that speed of DDR4 affects Skylakes gaming performance so now we need benches in "real life resolutions". For e.g. Anand only ran their dGPU tests with slowest DDR4, 2133MHz.

Anand hasn't been so hot (in my eyes) as far as reviews go in some time.

I think that test is a bit specific to the point where you'll have to wait until some people get the CPUs in their hands. I'm curious to also see how much DDR4 speeds affect non-FPS areas of gaming. Especially would like to see performance in MMOs or open-world games that could be streaming or caching a lot of assets.

 

[digitalrelic]

Resolutions are low because they want to kill any GPU bottleneck there could be and get pure CPU+RAM performance tested. Sure that difference doesn't translate 1:1 into e.g. 1440p gaming, but clearly indicates that Skylake performance scales upwards as speed of memory increases.

You try to remove the impact of the discrete GPU when testing the performance the CPU has on gaming.

The purpose of testing at low res is to remove any GPU bottleneck so the differences in CPU architecture become more apparent.

I understand all that but again, I don't see the point of benchmarks like these when they will have literally zero impact in real-world gaming performance.