Intel Core i5 11600K + Core i9 11900K Linux Performance Across ~400 Benchmarks

[coder]

Thanks for the benchmarks, Michael! I appreciate that you separated the Geo Means for the CPU and GPU tests, as well.

[coder]

well said ....All that Silicon could have been a nice anticipated GFX Chip or even a more usefull CPU instead.

Really? At 14 nm? What better than this are you going to do at 14 nm? A 14 nm GPU would be completely uncompetitive, today!

In my opinion, this is positive because it lets Intel use their existing 14 nm production capacity to help satiate the ravenous, world-wide demand for CPUs. And with a product that's objectively better than their previous gen.

Is it the CPU we all wish it were? Clearly not. But, I do think it's better than if they just kept selling Comet Lake until Alder Lake is ready. For one thing, it has more and faster PCIe lanes, so that's a help for some of us. The iGPU upgrades are also appreciated, not least of all AV1-support.

[coder]

QUOTE=angrypie;n1247908]20 Freekin' 23 !! By then AMD will be ...

I do love that enthusiasm.


I'm not counting Intel out, but also there's the ever-growing threat from ARM. Their v9 announcement is just in time to cast an ominous shadow over Intel's Ice Lake-SP parade.

[coder]

it's really a shame when you're fab limited to keep wasting half or more of the die for an iGPU that isn't needed. They're way past the iGPU that's helpful to just run a desktop doing normal web/text/spreadsheet kind of stuff. They're trying to get into the gaming space and that's just an uphill battle against dGPUs that they will never win.

A lot of people game on them. Just on the lowest settings and maybe not with the latest games. If they were much less capable, it'd be bad news for a ton of people, right now, who cannot get or afford dGPUs.

Another thing they're good for is GPU-compute. While not the order-of-magnitude improvement that dGPUs offer, iGPUs are at least comparable to the CPU cores and on a far smaller power budget. You can run intel_gpu_top and see for yourself, while running some GPU-compute benchmarks.

Also, they're good at video encoding and transcoding acceleration. Again, not on the same level as Nvidia GPUs, but very well for what they are.

[coder]

my 10850k maxes out at 125 watts. in everything i do at stock enforcement of 125 watts on the P1, is 4.8ghz. even compiling my 10850k hovers around 4.6-4.8ghz with P1 enforced to 125 watts.

Lol. What's your Tau and PL2? PL1 doesn't matter, if Tau isn't being enforced or is set super-long.

mutli, a 5900x is faster, but the 5900x isn't worth $550 for the cpu + motherboard investment to shave off like 10 minutes or so in compiling ungoogle chromium. and the 5950x i might as well go threadripper and gain the pci-e lanes at that price range but i still won't get anything beneficial for what i do.

So, you're really just arguing that pricing of faster options is simply unattractive? But that can change.

And no, there's a huge gulf in pricing between the 5950X and Threadripper. I'll grant that the former straddles the line between mainstream and HEDT, but the 3000-series Threadrippers went and widened the gap back up, with yet higher pricing.

i did think about getting a 5800x when it came out since its $450 but all i would gain at most is 5% average in gaming. worthless along with slower multi.

I don't understand why most people even bother upgrading for less than a 50% improvement. So, no argument here.

[coder]

Intel also has the deep link feature that enables them to use their integrated+discrete gpus. It will be interesting to see if their xpu acceleration will be able to take advantage of multiple GPUs on pcie4... limited to 8 lanes per GPU. So, I think we haven't seen the full Rocket Lake story until we also see their new discrete GPUs.

Multi-GPU is hard enough when they're symmetrical and connected with a dedicated link. I really have very little expectation that the small 32 EU iGPU in Rocke Lake is going to be worth the trouble to load-balance with a big gaming dGPU.

The most sensible way for games to harness iGPUs is something they could already do today: offload other things onto them, like physics, AI, 3D sound processing, etc. I'm not sure if any actually do that, but I wouldn't be surprised.

The most compelling case for load-balancing between an iGPU and dGPU is in Intel's mobile platform, where they have 96 EU Xe GPUs in Tigerlake and as a standalone chip, both using DDR4. That's at least fairly symmetrical, and the best-case speedup of 2x would actually be worth some effort.

[CochainComplex]

Really? At 14 nm? What better than this are you going to do at 14 nm? A 14 nm GPU would be completely uncompetitive, today!

In my opinion, this is positive because it lets Intel use their existing 14 nm production capacity to help satiate the ravenous, world-wide demand for CPUs. And with a product that's objectively better than their previous gen.

Is it the CPU we all wish it were? Clearly not. But, I do think it's better than if they just kept selling Comet Lake until Alder Lake is ready. For one thing, it has more and faster PCIe lanes, so that's a help for some of us. The iGPU upgrades are also appreciated, not least of all AV1-suppbottleneck

So those high purity monocrystalline wafers are not suitable for <14nm processes? I highly doubt that. Growing a monocrystalline structure per se demands highest purity or the crystal structure will not be monocrystalline. Any defect will crack up the perfect lattice.

Concerning the current shortage of GPU's ...that silicon will be better be used for GPU dies. Yes I know that silicon supply is not the (only) bottleneck....

edit: found that gamers nexus review of the 11900K https://www.youtube.com/watch?v=mxiuvQPL_qs

[MadCatX]

As unimpressive as the RKL is I feel like these Linux tests made it look less bad than the mostly Windows tests done so far. At least the GPU is a nice improvement and the CPU itself did okay in lightly threaded workloads.

[piotrj3]

Honestly 11600k and below (11400 and 11500) processors looks good from price-performance point of view, especially considering new Intel iGPU supports AV1 hardware decoding/encoding and it truly improved performance so you can use new intel CPUs without GPUs (in current GPU crysis it is really good). 11600k is similar in performance to 5600X, has great iGPU and is cheaper (in my country by 15%). 11400/11500 are really cheap and great 6 cores.

11700k is questionable choice.
Anything higher 11700k is waste of money and sand.

I think Intel shouldn't release anything above 11700k, because i9 line up embarrases itself, while 11400-11600 can take a big piece of cake on market until AMD release more cheap options like 5600 (without X), and desktop Zen 3 APUs.

[angrypie]

As unimpressive as the RKL is I feel like these Linux tests made it look less bad than the mostly Windows tests done so far. At least the GPU is a nice improvement and the CPU itself did okay in lightly threaded workloads.

The most bloated Linux distro is still 10 times lighter than mainstream Windows 10 and Linux is better at SMP. Same reason why Bulldozer also tends to perform better on Linux.