Intel 10th Gen Comet Lake thread

[Destination]

Gamers nexus just put a video out on the 10850k.
OCUK price on this chip puts is as the best performance per £ for gaming by considerable margin. compared to the 10900k and 10700k
Interesting that it looks like overclocking is totally redundant for games with this chip which is a great place to be as you get all the performance without any of the power and heat problems.
Just ordered the 38GL950G LG 38" ultrawide and cant wait for the 3080/3090 launch now to finish the build.

I think you are seeing what you want to see.
The gamersnexus video stated very different conclusions to what you have started.
He had previously states the 10700K is worthless as a concept, and the 10900K is vastly over priced.
He basically stated the 10850K chip is utterly redundant for gaming, as the 10600K is the best price/performance by a massive margin, where your chip at twice the price, barely manages a couple of % above it, in most tests.

 

[Journey]

What ever the new naming scheme will be the 'next generation' CPU's are hopefully going to be great from Intel, it is a real shame that we won't see them until the end of 2021 or early 2022.

the 10600K is the best price/performance by a massive margin,

I have to agree, the 10600K is a great CPU if you are nailing the 1080p/1440p high refresh builds and will be very popular with the new RTX 3080, the rest of them are pointless as they offer no longevity, as the money saved could easily be used on the 'next generation' CPU's in '21 '22 and you'll end up with better for performance for longer, for around the same cost.

 

[LePhuronn]

What ever the new naming scheme will be the 'next generation' CPU's are hopefully going to be great from Intel, it is a real shame that we won't see them until the end of 2021 or early 2022.

yeah, Alder Lake vs Zen 4 should be pretty juicy if both live up to the scant rumours out there.

 

[pc-guy]

It kind of doesnt matter though does it. The AMD CPU will wipe out the gains and then more (if there are any gains)

what on earth that is holy...

 

[Yaayuh!]

Gamers nexus just put a video out on the 10850k.

OCUK price on this chip puts is as the best performance per £ for gaming by considerable margin. compared to the 10900k and 10700k

Interesting that it looks like overclocking is totally redundant for games with this chip which is a great place to be as you get all the performance without any of the power and heat problems.

Just ordered the 38GL950G LG 38" ultrawide and cant wait for the 3080/3090 launch now to finish the build.

Why bother with the 38GL950G if you're gaming at 1080P or less?

 

[pc-guy]

Why bother with the 38GL950G if you're gaming at 1080P or less?

@fx1 is assuming that with 3080 or 3090 he is gonna be gaming @ 1440p high refreshrate. but forgetting that PCIe Gen 3 on his intel CPU will likely be a bottle neck for GPU performances...

https://youtu.be/PAwIh1nSOQ8

 

[fx1]

@fx1 is assuming that with 3080 or 3090 he is gonna be gaming @ 1440p high refreshrate. but forgetting that PCIe Gen 3 on his intel CPU will likely be a bottle neck for GPU performances...

https://youtu.be/PAwIh1nSOQ8

wtf are you on about. PCI3 is no bottleneck.

You clearly have no idea about the res of the LG either by looks of it.

 

[Journey]

WQHD+ 3840x1600 144Hz. So gaming at a native resolution with any CPU from any company won't get rid of any GPU bottleneck. Yes it's 25% less pixels than 4k, but you'll still be GPU limited, regardless of how you try to show otherwise.

wtf are you on about. PCI3 is no bottleneck.

Id' not be so sure until you see what the reviews show next week especially on the RTX 3090, or you can just continue to assume you know everything, right?

 

[fx1]

WQHD+ 3840x1600 144Hz. So gaming at a native resolution with any CPU from any company won't get rid of any GPU bottleneck. Yes it's 25% less pixels than 4k, but you'll still be GPU limited, regardless of how you try to show otherwise.



Id' not be so sure until you see what the reviews show next week especially on the RTX 3090, or you can just continue to assume you know everything, right?

Well im happy to be corrected but PCI3 will be more than enough i think for the 3090.

Ive also been around long enough to know that average frames in benchmarks dont tell the full story. Min and Max FPS and stuttering is just as important to me as average FPS. Games are not 100% GPU limited especially in multiplayer games where you can hit CPU and Memory bottlenecks whilst being overall limited by the GPU.

 

[Journey]

Min and Max FPS and stuttering is just as important to me as average FPS.

0.1% an 1% lows are indeed all that matter, and again any decent CPU even a lowly Intel 10400F would be fine and not cause stuttering at such high resolutions as there is such little demand on the CPU compared with 1080p, sure once you hit 200fps+ then the CPU will start to sweat maybe in some games, but there aren't many games you'll get 3840x1600 at 200+ fps.

 

[pc-guy]

wtf are you on about. PCI3 is no bottleneck.

You clearly have no idea about the res of the LG either by looks of it.

Just keep on living in your own universe. Good for you! I mean that level of ignorance and stubbornness...

basic logic for you based on the performance tested for pcie3 vs pcie4 on 5700xt:
5700xt @ pcie 3 < 2080ti @ pcie 3- eq 1
5700xt @ pcie3 < 5700xt @ pcie4 - eq 2
3080 @ pcie3 > 2080ti - eq 3 (benchmark demonstrates by nvidia is based on intel system therefore only pcie3 available)
3090 @ pcie3 > 2080ti - eq 4 (benchmark demonstrates by nvidia is based on intel system therefore only pcie3 available)

Now sub eq 3 & 4 into 1 and subsequently to 2 then you can derive the following

3080 pcie3 < 3080 pcie4
3090 pcie3 < 3090 pcie4

it wouldn’t be too far fetched to think that if the 2080ti is a pcie 4 card it may well be able to squeeze more frames out of it.

 

[fx1]

Just keep on living in your own universe. Good for you! I mean that level of ignorance and stubbornness...

basic logic for you based on the performance tested for pcie3 vs pcie4 on 5700xt:
5700xt @ pcie 3 < 2080ti @ pcie 3- eq 1
5700xt @ pcie3 < 5700xt @ pcie4 - eq 2
3080 @ pcie3 > 2080ti - eq 3 (benchmark demonstrates by nvidia is based on intel system therefore only pcie3 available)
3090 @ pcie3 > 2080ti - eq 4 (benchmark demonstrates by nvidia is based on intel system therefore only pcie3 available)

Now sub eq 3 & 4 into 1 and subsequently to 2 then you can derive the following

3080 pcie3 < 3080 pcie4
3090 pcie3 < 3090 pcie4

it wouldn’t be too far fetched to think that if the 2080ti is a pcie 4 card it may well be able to squeeze more frames out of it.

You realise that the difference between PCIe 3 8x and 16x is 2% with a 2080ti..... and 8x has 50% less bandwidth.

How much do you think you are going to get from PCIe4?

 

[Journey]

It's not about the sustained throughput, it is about the burst requirements through the bus from the card. There are peaks and troughs, and the peaks can sometimes hit a card hard, just like those 0.1% and 1% lows we were talking about earlier. Consider the bandwidth that the new 3090 has for its RAM, and then look at what is required to feed that, then you'll see why the peaks are the important part.

 

[fx1]

0.1% an 1% lows are indeed all that matter, and again any decent CPU even a lowly Intel 10400F would be fine and not cause stuttering at such high resolutions as there is such little demand on the CPU compared with 1080p, sure once you hit 200fps+ then the CPU will start to sweat maybe in some games, but there aren't many games you'll get 3840x1600 at 200+ fps.

You can build a computer with cheaper CPU's with less cores. That works fine until you have 4 cores and then a game needs 6. You buy a 6 core and then a game uses 8 and so on.

There are places to save money if you want to. Such as 10900k down to 10850k. My 6700k has lasted me 5 years and retained 50% of its value. I cant really grumble but with only 4 cores its showing its age and its overclock became unstable.

 

[fx1]

It's not about the sustained throughput, it is about the burst requirements through the bus from the card. There are peaks and troughs, and the peaks can sometimes hit a card hard, just like those 0.1% and 1% lows we were talking about earlier. Consider the bandwidth that the new 3090 has for its RAM, and then look at what is required to feed that, then you'll see why the peaks are the important part.

But they arent important. Nobody cares about losing 2 frames from the top of a peak because it affects nothing. You wouldnt even notice it.

On the other hand hitting a memory or CPU bottleneck can send your frames crashing which you will notice every time.

Id also argue that if a game spikes that high its most likely due to bad coding rather than genuine bandwidth requirement.

 

[Journey]

Your 6700K only lasted this long as the same 4c/8t CPU's were being sold from 2011 until 2017.

Hence why in the ever chaining world of MOAR CORES!!!!!! can you say that the 10850K at double the nearly cost of the 10600K will last as long, and if not what will if offer you in the first two years of ownership over the 10600K as you've put it the requirements from games will only go up.

The days of systems lasting 5+ years at peak performance are long gone, until we settle into another rut of development from the manufactures.

But they arent important. Nobody cares about losing 2 frames from the top of a peak because it affects nothing. You wouldnt even notice it.

A peak meaning the peak bandwidth requirement of the card, not the peak frame rate, so if the card misses the required data it has to slow down to wait for the data, which is when you get the 1% lows you don't want.

 

[smilingcrow]

A peak meaning the peak bandwidth requirement of the card, not the peak frame rate, so if the card misses the required data it has to slow down to wait for the data, which is when you get the 1% lows you don't want.

I'd wondered if PCIe 4.0 gives noticeably better frame rates at times even though the average rate so far hasn't gone up significantly.
Have you seen an article where they go into this in detail?

 

[Journey]

I'd wondered if PCIe 4.0 gives noticeably better frame rates at times even though the average rate so far hasn't gone up significantly.
Have you seen an article where they go into this in detail?

You'll see loads of data from next week showing this, since we'll finally get much faster consumer GPUs with 4.0 support

 

[smilingcrow]

You'll see loads of data from next week showing this, since we'll finally get much faster consumer GPUs with 4.0 support

Okay, I thought you were referring to tests that you had already seen on current cards.
The whole PCIe 4.0 issue may be more nuanced than some are predicting.
I don't know as haven't looked into the minutia of it.

 

[Journey]

Okay, I thought you were referring to tests that you had already seen on current cards.
The whole PCIe 4.0 issue may be more nuanced than some are predicting.
I don't know as haven't looked into the minutia of it.

You can see it now just not using games, you need to be using other workloads, but people on this forum only care about games.

https://www.igorslab.de/en/pci-expr...al-workloads-compared-and-some-wow-moments/3/