8700K vs 2700X on with 2080Ti [computerbase]

[gdansk]

Hardware Unboxed's video on the same subject:

As he says, there will be more when they can include the 9900K.

 

[Arkaign]

Hardware Unboxed's video on the same subject:

As he says, there will be more when they can include the 9900K.

Definitely a good apples to apples comparison, same ram/XMP is important.

I am running 3733 Patriot at CL15 on my 8086k/Aorus 5. I have seen decent gains in some titles over my old 3200 kit.

I kind of wonder about his OC opinion though. Stock 8700k doesn't do much at 6C. But with good cooling and all-core @4.8+ you see some pretty sizable gains. I can do 5.0 6C, but beyond that I probably need better cooling and/or delid.

2700x remains a terrific value. 8700k remains (for the next few days at least) the best pure gaming option. With people keeping the same CPU through multiple GPU revisions, we can already see the gap widening vs 1080ti results with the same CPUs. Whereas 1440p was a near tie before, now a gap is really showing up, that one title being ~80fps vs ~100fps at 1440 should be interesting for high refresh gamers. The ~2% gap at 4k means that there really is no advantage possible there, GPU wall with 2080 at those settings.

In my case, I run 3440x1440 UW Gsync. And I do *not* like to max out every setting. I dislike most AA, and don't mind dropping a couple of minor settings to keep a rock solid 100+ as possible. Someone like me sees results closer to 1080/max settings than 1440/max settings, so CPU is a bottleneck I absolutely want to avoid.

Someone with a 1080ti running 1440 or higher all Ultra+AA almost certainly would see basically zero difference between a 2700x and 8700k. Ditto 4k/60hz panels running even a 2080ti.

However, I think the 1440p gap shown here will become the gap we see at 4k with the 7nm Nvidia cards. 10xx was a really really long lived line. I don't think 20xx will see nearly the same length before replacement. 12nm is going to be way fewer dies per wafer than 7nm or 5nm provide, so I honestly feel like this is more of a placeholder product until the big swing happens probably next fall. The Ti launching simultaneously gives me even more of that impression.

 

[epsilon84]

Definitely a good apples to apples comparison, same ram/XMP is important.

I am running 3733 Patriot at CL15 on my 8086k/Aorus 5. I have seen decent gains in some titles over my old 3200 kit.

I kind of wonder about his OC opinion though. Stock 8700k doesn't do much at 6C. But with good cooling and all-core @4.8+ you see some pretty sizable gains. I can do 5.0 6C, but beyond that I probably need better cooling and/or delid.

2700x remains a terrific value. 8700k remains (for the next few days at least) the best pure gaming option. With people keeping the same CPU through multiple GPU revisions, we can already see the gap widening vs 1080ti results with the same CPUs. Whereas 1440p was a near tie before, now a gap is really showing up, that one title being ~80fps vs ~100fps at 1440 should be interesting for high refresh gamers. The ~2% gap at 4k means that there really is no advantage possible there, GPU wall with 2080 at those settings.

In my case, I run 3440x1440 UW Gsync. And I do *not* like to max out every setting. I dislike most AA, and don't mind dropping a couple of minor settings to keep a rock solid 100+ as possible. Someone like me sees results closer to 1080/max settings than 1440/max settings, so CPU is a bottleneck I absolutely want to avoid.

Someone with a 1080ti running 1440 or higher all Ultra+AA almost certainly would see basically zero difference between a 2700x and 8700k. Ditto 4k/60hz panels running even a 2080ti.

However, I think the 1440p gap shown here will become the gap we see at 4k with the 7nm Nvidia cards. 10xx was a really really long lived line. I don't think 20xx will see nearly the same length before replacement. 12nm is going to be way fewer dies per wafer than 7nm or 5nm provide, so I honestly feel like this is more of a placeholder product until the big swing happens probably next fall. The Ti launching simultaneously gives me even more of that impression.

Good post, I basically agree with everything you said there.

With a 2080 Ti, an overclocked 8700K *should* start to pull away from the 2700X, even at 1440P. Even at stock, there is about an 8% gap there whereas with the 1080 Ti it was basically non existent as you said. This is of course most relevant to 100Hz+ panel gamers. 60Hz gamers will probably be better off upgrading the monitor more than anything else!

Also agree about not maxing out every possible setting in game to ensure higher and more stable framerates. Most of the time the visual difference between 'ultra' and 'high' isn't that great but the performance hit can be very substantial.

 

[happy medium]

We always hear the argument, buy the AMD platform because you can upgrade till the year 2020.
Ok so if you bought a 1700, and then upgraded to a 2700 or skipped the 2700 and went straight to a 3700 cpu will the 3700 CPU beat a 8700k at 5.0ghz?
I think it will be real close.
Is it worth buying 2 or 3 cpu's for the price of one 8700k?
Add the price of a 1700 and a 2700 or a 1700 and a 3700 next year, your paying more for less performance.
And you could have been rocking the top performance of a 8700k at 5.0 for 3 years by then.

By next year a 1700 CPU will be bottlenecking rtx3060 and will bottleneck a rtx2070 this year unless your gaming at 4k.

Next year a 2700 will be bottlnecking a rtx3070 at 1440p just like it does with a rtx2080ti now.

 

[epsilon84]

We always hear the argument, buy the AMD platform because you can upgrade till the year 2020.
Ok so if you bought a 1700, and then upgraded to a 2700 or skipped the 2700 and went straight to a 3700 cpu will the 3700 CPU beat a 8700k at 5.0ghz?
I think it will be real close.
Is it worth buying 2 or 3 cpu's for the price of one 8700k?
Add the price of a 1700 and a 2700 or a 1700 and a 3700 next year, your paying more for less performance.
And you could have been rocking the top performance of a 8700k at 5.0 for 3 years by then.

By next year a 1700 CPU will be bottlenecking rtx3060 and will bottleneck a rtx2070 this year unless your gaming at 4k.

Next year a 2700 will be bottlnecking a rtx3070 at 1440p just like it does with a rtx2080ti now.

From a gaming perspective, those are fair points. You can always sell the older AMD processor of course to offset the cost of the new CPU, but that is an additional hassle.

I too have my doubts that Zen 2 can trump my current 8700K @ 5.0GHz for gaming purposes. It's one thing to come close, it's another to have one significantly faster to convince me to upgrade. Basically a overclocked 2700X @ 4.2GHz performs like an i5 8400 in games... yes, thats a 8C/16T Ryzen @ 4.2GHz performing like a 6C/6T i5 @ 3.8GHz for games. And people expect AMD's next chip to suddenly best a 6C/12T i7 @ 5.0GHz?? Call me a pessimist or whatever, but I'll believe it when I see it.

A lot of people are expecting Zen 2 to come with (IMO) unrealistic IPC improvements and clocks, but I'll be happy to be proven wrong and have AMD pull another Athlon 64 type improvement out of the hat. Until proven though, it's all guesswork as to just how good Zen 2 can be.

To be fair, I doubt anything Intel has at 10nm will be drastically faster either. At least not until the 10nm refresh, or 10nm++ or whatever. So we're talking 2020 and beyond. So I think I'll stick with my 8700K for a while... unless I sell it now while there is a CPU shortage, take my vacation (will be gone for 2 months) and then buy a 9900K once I get back and prices and supplies settle in the new year. Win win!

 

[Midwayman]

2700x remains a terrific value. 8700k remains (for the next few days at least) the best pure gaming option. With people keeping the same CPU through multiple GPU revisions, we can already see the gap widening vs 1080ti results with the same CPUs. Whereas 1440p was a near tie before, now a gap is really showing up, that one title being ~80fps vs ~100fps at 1440 should be interesting for high refresh gamers. The ~2% gap at 4k means that there really is no advantage possible there, GPU wall with 2080 at those settings.

In my case, I run 3440x1440 UW Gsync. And I do *not* like to max out every setting. I dislike most AA, and don't mind dropping a couple of minor settings to keep a rock solid 100+ as possible. Someone like me sees results closer to 1080/max settings than 1440/max settings, so CPU is a bottleneck I absolutely want to avoid.

That's always been my feeling. I keep a cpu through several GPU generations. I'll turn down setting to keep minimums high. Unless you're stuck with 60hz panels ultra testing at 1440k and 4k has never made a lot of sense to me.

 

[Glo.]

Update on the topic:

"Our test results were wrong".

https://www.computerbase.de/2018-09/ryzen-7-2700x-core-i7-8700k-geforce-rtx-2080-ti/


I wonder what the author of this thread now will say .

 

[caswow]

There are others looking silly right now too

 

[Hitman928]

Update on the topic:

"Our test results were wrong".

https://www.computerbase.de/2018-09/ryzen-7-2700x-core-i7-8700k-geforce-rtx-2080-ti/


I wonder what the author of this thread now will say .

They said a fresh install of Windows is what "cured" the performance problem on Ryzen. It sounds like they were using Windows images to quickly get each system up and running and probably used a Windows image that was configured for intel on the AMD system as well. Just a guess, but if everything else was checked and a fresh install solved the problem, it seems a likely candidate.

 

[Glo.]

They said a fresh install of Windows is what "cured" the performance problem on Ryzen. It sounds like they were using Windows images to quickly get each system up and running and probably used a Windows image that was configured for intel on the AMD system as well. Just a guess, but if everything else was checked and a fresh install solved the problem, it seems a likely candidate.

Apparently, that is exactly what has happened here.

Aren't you always supposed to do a clean instal if you are changing your platform, between vendors?

 

[Hitman928]

Apparently, that is exactly what has happened here.

Aren't you always supposed to do a clean instal if you are changing your platform, between vendors?

That is best practice, yes.

 

[Kenmitch]

There are others looking silly right now too

Everybody looks silly to somebody. Cyber silly is less offensive to most people. Guess for some being not cyber silly is all that's left.

 

[Muhammed]

Update on the topic:

"Our test results were wrong".

https://www.computerbase.de/2018-09/ryzen-7-2700x-core-i7-8700k-geforce-rtx-2080-ti/


I wonder what the author of this thread now will say .

I say fair is fair. Though 17% is still a significant difference IMO.

 

[Gikaseixas]

I say fair is fair. Though 17% is still a significant difference IMO.

you should edit OP. Its now half of what was claimed to be

 

[Muhammed]

The 9900K is 26% faster than 2700X according to Computerbase
https://www.computerbase.de/2018-10...t/2/#abschnitt_benchmarks_in_spielen_fhd__uhd

 

[Malogeek]

The 9900K is 26% faster than 2700X according to Computerbase
https://www.computerbase.de/2018-10...t/2/#abschnitt_benchmarks_in_spielen_fhd__uhd

One would hope so for double the price!

 

[moinmoin]

The 9900K is 26% faster than 2700X according to Computerbase
https://www.computerbase.de/2018-10...t/2/#abschnitt_benchmarks_in_spielen_fhd__uhd

While consuming up to 37% more watts. Simple brute force performance.

 

[jpiniero]

While consuming up to 37% more watts. Simple brute force performance.

That extra hundred mhz or two costs a ton in power if you have to raise voltage a lot, especially x8.

 

[TheELF]

While consuming up to 37% more watts. Simple brute force performance.

Lol,seriously?What else is there?Also there is underclocking/undervolting for anybody who want's lower consumption,you can power off cores you don't need and so on.

 

[zinfamous]

While consuming up to 37% more watts.

and 100% more dollars

 

[JPB]

We always hear the argument, buy the AMD platform because you can upgrade till the year 2020.
Ok so if you bought a 1700, and then upgraded to a 2700 or skipped the 2700 and went straight to a 3700 cpu will the 3700 CPU beat a 8700k at 5.0ghz?
I think it will be real close.
Is it worth buying 2 or 3 cpu's for the price of one 8700k?
Add the price of a 1700 and a 2700 or a 1700 and a 3700 next year, your paying more for less performance.
And you could have been rocking the top performance of a 8700k at 5.0 for 3 years by then.

By next year a 1700 CPU will be bottlenecking rtx3060 and will bottleneck a rtx2070 this year unless your gaming at 4k.

Next year a 2700 will be bottlnecking a rtx3070 at 1440p just like it does with a rtx2080ti now.

If Zen 2 ends up with a 13% IPC lift, then I expect Zen 3 will have "zero" problem of probably beating a 8700K at 5Ghz....And yes, the OP needs to update the original post to better reflect the benchmark updates.

AMD's next-gen Zen 2: 13% IPC improvement, better perf soon

Read more: https://www.tweaktown.com/news/6355...3-ipc-improvement-better-perf-soon/index.html

 

[.vodka]

We always hear the argument, buy the AMD platform because you can upgrade till the year 2020.
Ok so if you bought a 1700, and then upgraded to a 2700 or skipped the 2700 and went straight to a 3700 cpu will the 3700 CPU beat a 8700k at 5.0ghz?
I think it will be real close.
Is it worth buying 2 or 3 cpu's for the price of one 8700k?
Add the price of a 1700 and a 2700 or a 1700 and a 3700 next year, your paying more for less performance.
And you could have been rocking the top performance of a 8700k at 5.0 for 3 years by then.

By next year a 1700 CPU will be bottlenecking rtx3060 and will bottleneck a rtx2070 this year unless your gaming at 4k.

Next year a 2700 will be bottlnecking a rtx3070 at 1440p just like it does with a rtx2080ti now.

Why would you upgrade from a 1700 to a 2700/2700x? It's a marginal performance increase from a 1700 @ 3.8-3.9GHz and 3200-3466MHz memory with proper timings... not too far behind that 5GHz 8700k, too. The value proposition is there for the smart buyer, and the real upgrade path is a 3xxx or 4xxx series CPU from a 1xxx part, not upgrading every generation.

This will probably be different for who bought into AM4 and a 2xxx series CPU, since next gen features a proper node shrink and a second generation architecture with the possibility of great double digit improvements on average everywhere. They might be tempted enough to get a 3xxx CPU.

Of course if you want to upgrade the CPU every generation then that's fine, but it's not smart for your wallet.

If you wanted to do this with an Intel platform you'd have to throw everything away in less than two years. These past few years, with every Skylake iteration thanks to the 10nm disaster, it's been a dumpster fire with copy pasted chipsets and socket revisions... Icelake will feature yet again a new chipset and a new socket. I guess you get to keep your DDR4... if they don't adopt DDR5 by 2020, that is.

Meanwhile AMD will go to a new socket when DDR5 and PCIe4/5 is ready, or they'll probably do an AM4+ mid cycle upgrade with backwards compatibility as usual, if needed. They didn't unify AM1/FM2+/AM3+ to AM4 for nothing, you know.

 

[Malogeek]

Why would you upgrade from a 1700 to a 2700/2700x? It's a marginal performance increase from a 1700 @ 3.8-3.9GHz and 3200-3466MHz memory with proper timings... not too far behind that 5GHz 8700k, too. The value proposition is there for the smart buyer, and the real upgrade path is a 3xxx or 4xxx series CPU from a 1xxx part, not upgrading every generation.

I upgraded from my 1700 to a 2600X, losing 2c/4t in the process and never been happier with my PC. I could not run my 1700 at 3.7Ghz forced with any kind of stability, nor my Samsung-b memory over 2933. Had it for a year and hated the damn thing, apart from Handbrake. Leaving it to boost itself was about 3.2Ghz. My 2600X boosts itself out-of-the-box to 4.2Ghz in games and 4Ghz in handbrake sustained and my RAM is now running great with tight timings. Didn't need to mess with overclocking at all.

The difference between 6c/12t and 8c/16t for 95% of my use is negligible (games). Also selling my 1700 and getting 2600X on sale meant a difference of $60 total. It's not something I'd suggest to everyone but it's most certainly an upgrade in a lot of cases.

 

[TheELF]

If Zen 2 ends up with a 13% IPC lift, then I expect Zen 3 will have "zero" problem of probably beating a 8700K at 5Ghz....And yes, the OP needs to update the original post to better reflect the benchmark updates.

AMD's next-gen Zen 2: 13% IPC improvement, better perf soon

Read more: https://www.tweaktown.com/news/6355...3-ipc-improvement-better-perf-soon/index.html

"13% average increase in IPC performance in "scientific tasks" with "no gaming data" provided right now"
So it will get better at 7zip or handbrake but not games.

 

[PotatoWithEarsOnSide]

You can guarantee that they'll have targeted reducing latency as a priority, so gaming will undoubtedly improve. The question is how much of that IPC improvement is from improved latencies, since Zen+ gained more in gaming performance than the improvement in latency.