Speculation: i9-9900K is Intel's last hurrah in gaming

[Despoiler]

https://www.gamersnexus.net/hwrevie...paste-delid-gaming-benchmarks-vs-2700x/page-4

With a 2080 Ti, the 9900K is opening up a serious performance gap on the 2700X, as expected.

Wonder if the OP still thinks this kind of gap can be bridged, let alone surpassed, with Zen 2??

If you want your power draw to exceed your stated TDP by 150%.....sure. Amazing gap they have.

https://www.anandtech.com/show/13400/intel-9th-gen-core-i9-9900k-i7-9700k-i5-9600k-review/21

 

[epsilon84]

If you want your power draw to exceed your stated TDP by 150%.....sure. Amazing gap they have.

https://www.anandtech.com/show/13400/intel-9th-gen-core-i9-9900k-i7-9700k-i5-9600k-review/21

I don't really care about TDP to be honest... I knew all along the 95W figure was for the 'base clock' and actual power consumption would be much higher (check my previous posts on this topic if you wish)

I think the 46% lead it has over an overclocked 2700X in FC5 is pretty impressive though, irrespective of the power draw:

As is the 33% margin over the 2700X in AC:O

 

[french toast]

Who the hell is going to play a game at 1080p normal with a 2080ti?
That is a worst case scenario to say the least.

 

[epsilon84]

Who the hell is going to play a game at 1080p normal with a 2080ti?
That is a worst case scenario to say the least.

I am referring to the outright lead the 9900K has over a 2700X in 'CPU bound' gaming.

This is the gap that the OP thinks Zen 2 will be able to bridge. Based on these results, I find that highly improbable.

1440P results aren't all that different:

Who would play at reduced settings? Well, maybe people with 144Hz monitors that actually want an average framerate close to the monitor refresh rate.

I personally usually play at 'high' settings instead of 'ultra' because I find the difference in visual quality to be small, but ultra quality usually means I fall below 60fps too often in games, which is a big no no for me.

 

[coercitiv]

I am referring to the outright lead the 9900K has over a 2700X in 'CPU bound' gaming.

This is the gap that the OP thinks Zen 2 will be able to bridge. Based on these results, I find that highly improbable.

A bridge too close, a bridge too far, whatever the wind brings.

Wouldn't you say we are already there in terms of AMD being 'close enough' in gaming? A 2700X is within 10% of a 8700K at 1080P. The 9900K will extend that lead a bit but at a signicantly higher cost. Zen 2 will reduce that margin back under 10%

 

[Despoiler]

I don't really care about TDP to be honest... I knew all along the 95W figure was for the 'base clock' and actual power consumption would be much higher (check my previous posts on this topic if you wish)

I think the 46% lead it has over an overclocked 2700X in FC5 is pretty impressive though, irrespective of the power draw:

As is the 33% margin over the 2700X in AC:O

Intel apparently doesn't care about TDP either. Granted TDP and power draw are NOT the same, but generally speaking they are ball park to each other. 150% isn't ball park. At that point you are forced to go under custom water to guarantee you are maximizing the chip because the alternative is throttling. You should be anyways if you are spending top dollar on a rig, but custom loops are still surprisingly uncommon.

I am referring to the outright lead the 9900K has over a 2700X in 'CPU bound' gaming.

This is the gap that the OP thinks Zen 2 will be able to bridge. Based on these results, I find that highly improbable.

1440P results aren't all that different:

Who would play at reduced settings? Well, maybe people with 144Hz monitors that actually want an average framerate close to the monitor refresh rate.

I personally usually play at 'high' settings instead of 'ultra' because I find the difference in visual quality to be small, but ultra quality usually means I fall below 60fps too often in games, which is a big no no for me.

Assuming 1440p you could have already had that with an OC 8700k. It's also cheaper overall.

I don't know about last hurrah vs AMD, but it certainly looks like it's the last hurrah for Intel's 14nm.

 

[epsilon84]

Intel apparently doesn't care about TDP either. Granted TDP and power draw are NOT the same, but generally speaking they are ball park to each other. 150% isn't ball park. At that point you are forced to go under custom water to guarantee you are maximizing the chip because the alternative is throttling. You should be anyways if you are spending top dollar on a rig, but custom loops are still surprisingly uncommon.

Assuming 1440p you could have already had that with an OC 8700k. It's also cheaper overall.

I don't know about last hurrah vs AMD, but it certainly looks like it's the last hurrah for Intel's 14nm.

Actually, as a gaming CPU, a 9900K wouldn't draw *that* much more than a 8700K, with both overclocked to the same clocks, at the same voltage - that is because no game can come close to maxing out the available cores/threads on a 9900K. Gaming isn't something that will peg all 8 cores / 16 threads under full load, it won't even come close. Even a 8700K isn't close to 100% CPU utilisation in gaming.

All this hoopla about high power consumption and temps doesn't really translate into gaming - I can verify this with my own overclocked 8700K. At 5.0GHz temps can reach into the high 80s with stress testing (I only have a CM 212), but since I'm mainly a gamer, load temps when gaming are usually in the 60s which is perfectly fine by me.

Couldn't agree more that it's the last hurrah for 14nm... Intel is definitely hitting the power / efficiency wall by clocking the 9900K so aggresively.

 

[french toast]

"
"The 9900K isn’t really a direct comparison with the 2700 ($250) or 2700X, anyway. The price disparity is huge. The 2700 can be had for significantly cheaper, then overclocked for still-good gaming performance, while leaving more budget for better GPUs or RAM.

The performance gap looks a lot worse for AMD when summed-up as a percentage, but in reality, as illustrated above, that percentage is potentially misleading. Statistics can be funny in that way. We’re talking differences of 3.7ms frametimes vs. 4.7ms frametimes, in some of these games, and the level to which most people can detect that delta is questionable. One could argue that the faster CPU will “age better,” but that’s a tough argument – the usable life of the system may expire before that argument matters."- gamers Nexus.

There is the issue, the 9900k is not that much faster at 1440p gaming overall, which is what most people will be playing with, let alone 4k.
Even in 1080p medium with a crazy fast 2080ti in a worst case scenario CPU bound game for AMD...you are looking at 3.7ms Vs 4.7ms...no one can tell the difference between them....by the time you might see the difference in a few years time it will be time to upgrade anyway.

9900k and 9700k make no sense what so ever, honestly just buy an 8700k for gaming or overclock a 2700 for best all round value.
Too hot and too expensive.

 

[LTC8K6]

So if it is shrunk to the 10nm+ process, it should be a real beast?

 

[epsilon84]

"
"The 9900K isn’t really a direct comparison with the 2700 ($250) or 2700X, anyway. The price disparity is huge. The 2700 can be had for significantly cheaper, then overclocked for still-good gaming performance, while leaving more budget for better GPUs or RAM.

The performance gap looks a lot worse for AMD when summed-up as a percentage, but in reality, as illustrated above, that percentage is potentially misleading. Statistics can be funny in that way. We’re talking differences of 3.7ms frametimes vs. 4.7ms frametimes, in some of these games, and the level to which most people can detect that delta is questionable. One could argue that the faster CPU will “age better,” but that’s a tough argument – the usable life of the system may expire before that argument matters."- gamers Nexus.

There is the issue, the 9900k is not that much faster at 1440p gaming overall, which is what most people will be playing with, let alone 4k.
Even in 1080p medium with a crazy fast 2080ti in a worst case scenario CPU bound game for AMD...you are looking at 3.7ms Vs 4.7ms...no one can tell the difference between them....by the time you might see the difference in a few years time it will be time to upgrade anyway.

9900k and 9700k make no sense what so ever, honestly just buy an 8700k for gaming or overclock a 2700 for best all round value.
Too hot and too expensive.

Since when did this become about value? If you want value, don't game on a 2080 Ti period! It costs $1000+ last I checked?! If you are going to game on a 2080 Ti, chances are price/performance is more of an afterthought. You're after outright performance. So matching a 2080 Ti with a 2700X seems a bit odd to me, at the very least you should get an 8700K like you said, which isn't that far off a 9900K, unlike the 2700X.

As for bang for your buck, why stop at a 2700 though? A 2600 is $100 cheaper and would actually be my pick if I was overclocking on a budget. In fact that is the CPU I recommended to my cousin for his $1000 gaming build, its barely any slower than a Ryzen 7 and he can put that $100 towards a more powerful GPU = win win.

 

[french toast]

Since when did this become about value? If you want value, don't game on a 2080 Ti period! It costs $1000+ last I checked?! If you are going to game on a 2080 Ti, chances are price/performance is more of an afterthought. You're after outright performance. So matching a 2080 Ti with a 2700X seems a bit odd to me, at the very least you should get an 8700K like you said, which isn't that far off a 9900K, unlike the 2700X.

As for bang for your buck, why stop at a 2700 though? A 2600 is $100 cheaper and would actually be my pick if I was overclocking on a budget. In fact that is the CPU I recommended to my cousin for his $1000 gaming build, its barely any slower than a Ryzen 7 and he can put that $100 towards a more powerful GPU = win win.

No I mentioned 8700k..offers practically the same top end performance for much less money and without the crazy cooling requirements at stock.
I said the 2700 OC is a much better all round value proposition, but you are right that is besides the point.

 

[epsilon84]

No I mentioned 8700k..offers practically the same top end performance for much less money and without the crazy cooling requirements at stock.
I said the 2700 OC is a much better all round value proposition, but you are right that is besides the point.

As a 8700K owner, I would agree with that point, since I am running at 5.0GHz on air with a modest CM 212+ HSF. That probably ain't gonna happen on a 9900K.

You can get away with a good HSF for a stock 9900K though, especially if gaming is the most taxing thing you do on your system. As I said, no game will push a 9900K anywhere close to 100% usage. I'd be surprised to even see 50% usage to be honest - IIRC Ryzen 7 chips usually hover around the 30 - 40% mark in most games, and I don't see the 9900K being much different, seeing they are both 8C/16T CPUs.

 

[Timorous]

So looking at TPUs summary page (https://www.techpowerup.com/reviews/Intel/Core_i9_9900K/19.html) the 2700X at stock has 84.9% of the gaming performance as the 9900K at stock at 720p across their gaming suite.

So it looks like AMD need around a 17% uplift between IPC and raw clockspeed to match the 9900K on average in gaming. I think that is doable if the reported 13% IPC in scientific workloads is both true and applicable to gaming. I also expect they can do it with a much lower power draw.

 

[french toast]

So looking at TPUs summary page (https://www.techpowerup.com/reviews/Intel/Core_i9_9900K/19.html) the 2700X at stock has 84.9% of the gaming performance as the 9900K at stock at 720p across their gaming suite.

So it looks like AMD need around a 17% uplift between IPC and raw clockspeed to match the 9900K on average in gaming. I think that is doable if the reported 13% IPC in scientific workloads is both true and applicable to gaming. I also expect they can do it with a much lower power draw.

Yea I somewhat agree, whilst I am on record as saying I don't think ryzen 3000 will best a 9900k in gaming, I think it will close the gap to a point that they are virtually equal.
This with marginally better office performance and much better power consumption/heat/price.
Basically gaming parity down the stack.

 

[PotatoWithEarsOnSide]

As good as it is, I don't see it selling well in comparison to past halo products.
In all honesty, how many people are going to buy a CPU that needs custom water cooling and the absolute best motherboard for it to even function fully?
Sure, price is no object for many, but dicking about just to get it performing marginally better than the 9600k in gaming...? No thanks.
IMO, its definitely the nail in the coffin of 14nm+++++++++++++++++. Intel have pushed the node beyond where it could efficiently go.
These 9th gen CPUs will be the best Intel CPUs we see before DDR5 and PCIe5 are available.
If they cannot have something else ready by the time we get those, then that is where AMD will really gain that traction they really need.

 

[arandomguy]

So looking at TPUs summary page (https://www.techpowerup.com/reviews/Intel/Core_i9_9900K/19.html) the 2700X at stock has 84.9% of the gaming performance as the 9900K at stock at 720p across their gaming suite.

So it looks like AMD need around a 17% uplift between IPC and raw clockspeed to match the 9900K on average in gaming. I think that is doable if the reported 13% IPC in scientific workloads is both true and applicable to gaming. I also expect they can do it with a much lower power draw.

It's not simple because the scaling isn't perfect. Look at the numbers you're using. 9900k @ 5.1ghz is at least a 8.5% clock speed increase over their stock sample (possibly higher) but only represented a 3.9% increase in performance.

Gaming performance at the numbers we're looking at start to become limited in other ways as well such as memory performance. In some of the games tested at 720 the 9900k@5.1ghz has no performance gain over the stock 9900k or even 8700k.

 

[DrMrLordX]

As good as it is, I don't see it selling well in comparison to past halo products.

I don't see it selling well because it's out of stock everywhere, and may be for awhile . . .

 

[dark zero]

Yea I somewhat agree, whilst I am on record as saying I don't think ryzen 3000 will best a 9900k in gaming, I think it will close the gap to a point that they are virtually equal.
This with marginally better office performance and much better power consumption/heat/price.
Basically gaming parity down the stack.

They will close the gap, but the cost will kill the 9900K big time.

 

[french toast]

They will close the gap, but the cost will kill the 9900K big time.

Yea this is the end of the line for 14nm and skylake, even if they relax the process further they will always run into heat and power issues, they can only adjust slightly and drop prices.

Ryzen 3000 will perform similarly but at much better power and heat, affordable prices.

 

[Timorous]

It's not simple because the scaling isn't perfect. Look at the numbers you're using. 9900k @ 5.1ghz is at least a 8.5% clock speed increase over their stock sample (possibly higher) but only represented a 3.9% increase in performance.

Gaming performance at the numbers we're looking at start to become limited in other ways as well such as memory performance. In some of the games tested at 720 the 9900k@5.1ghz has no performance gain over the stock 9900k or even 8700k.

It is very simple. For Zen2 to match the 9900K in that suite of games, at those settings with that hardware it needs to increase performance by 17% on average.

If you change ram, motherboard, drivers, GPU, games, resolution, settings, windows patch version, bios version etc then it can change the result and therefore the required uplift.

I think that it shows that Zen2 matching the 9900k core for core and thread for thread is entirely doable and that is all it was ever meant to show.

Can AMD match Intel? Yes, it looks possible. Will they? Let's wait and see.

 

[jpiniero]

Yea this is the end of the line for 14nm and skylake, even if they relax the process further they will always run into heat and power issues, they can only adjust slightly and drop prices.

There's going to be one more 14 nm desktop chip (Coffee Lake Refresh Refresh). What Intel does to it, and if you would need a new socket is of course TBD. I'd have to say most likely it's just an effective price cut with the i7 getting HT and maybe the i5 and i3 with slightly higher stock clocks.

Beyond that, it's too early. I do wonder if Intel has focused all of their efforts on the Rapids; and whether the future Lakes is going to be minimal updates, but there should be avenues to improve efficiency.

 

[LTC8K6]

There's going to be one more 14 nm desktop chip (Coffee Lake Refresh Refresh). What Intel does to it, and if you would need a new socket is of course TBD. I'd have to say most likely it's just an effective price cut with the i7 getting HT and maybe the i5 and i3 with slightly higher stock clocks.

Beyond that, it's too early. I do wonder if Intel has focused all of their efforts on the Rapids; and whether the future Lakes is going to be minimal updates, but there should be avenues to improve efficiency.

I can't see how there is going to be one more gen of Intel 14nm desktop chips.
This is right at the limit. There's no more headroom for any better desktop chips.

CFL refresh refresh needs to be on 10nm+...

What happened to all those in-between 8000 series chips listed on Wikichip? XX20/XX50/XX70

https://en.wikichip.org/wiki/intel/microarchitectures/coffee_lake#All_Coffee_Lake_Chips

https://www.pcgamer.com/intels-unreleased-core-i7-8670-coffee-lake-shows-up-on-benchmarking-site/

 

[jpiniero]

I can't see how there is going to be one more gen of Intel 14nm desktop chips.
This is right at the limit. There's no more headroom for any better desktop chips.

There's probably some room to improve the low core count turbo. Intel said they are doing another transistor quality upgrade with Cooper Lake; presumably that applies to Comet and Coffee Refresh Refresh. I think it will still be branded 9th gen; just with the clocks altered and maybe adding HT.

As for the 8000 series chips, I think they changed their mind and went with 9000 series when they realized that Icelake wasn't happening on the desktop.

 

[Vattila]

Ryzen doesn't have any inherent weakness in min fps though […] It is simply a slower uarch for gaming, due to IPC and frequency deficits. Increasing either (or both, as will be the case of Zen 2) will bring AMD closer (or exceed, according to the OP) Intel levels of gaming performance. The big question is just how much of an improvement will we see? Will be it enough to break Intel's stranglehold in gaming, which has existed for over a decade since the launch of Core 2?I doubt it will, but the poll results and certain forum posters show a lot of optimism for AMD

Before I comment on i9-9900K, let me just reply to that last statement of yours about optimism:

I expect a lot from AMD based on competitive analysis, AMD's stated business plans and roadmap. As I have outlined before here, their plans would have to be ambitious to make sense. High-performance compute is highly competitive and failure is costly (ref. near bankruptcy in Bulldozer years). AMD cannot afford to fail.

My optimism stems from AMD CEO Lisa Su's excellent leadership and AMD's general execution since her decision to laser focus on the return to high-performance compute — a decision she took despite the risks involved and the recent near-death experience for the company. She could have taken the company in other directions. She did not. Show me signs that she is falling flat on her targets, and my optimism will wane pretty quickly. So far she has done well.

You will have a hard time lowering my expectations and convincing me that AMD is following a business plan that makes no sense, i.e. that their plans for Zen 2 has always been to lose against Ice Lake, or to just compete in a niche in the server market (which was the old Rory Read strategy with ARM), or to merely be a second-best value option in the mainstream (Lisa Su explicitly abandoned that losing strategy). Lisa Su aims for AMD to be a leader in high-margin high-performance CPU and GPU compute, especially where the combination of the two makes sense (such as HPC server, gaming, notebook, embedded and semi-custom design).

I am referring to the outright lead the 9900K has over a 2700X in 'CPU bound' gaming. This is the gap that the OP thinks Zen 2 will be able to bridge.

We have had a lot of good replies and discussion in this thread. I fully concede that there will be outliers that AMD may not win. However, they need to win on modern code. That includes modern well-written game code. There is no subset of the x86 ISA that is particular to gaming. There is however a lot of legacy stuff that AMD may be wise to consider less important. But on modern code designed with industry best practices and built with modern compilers, AMD needs to take the lead with Zen 2.

Again, Zen 2 would have had to be planned to match Ice Lake level of performance. The latter is not here yet due to Intel's stumbles. AMD now has an unprecedented opportunity, and AMD executives have stated as much.

The i9-9900K is an impressive last hurrah on 14nm (kudos to Intel engineers!). Perhaps it even is at a performance level originally projected for the initial 10nm Ice Lake. Even so, we should all expect 7nm Zen 2 to beat it — unless something has gone wrong at AMD.

So if it is shrunk to the 10nm+ process, [i9-9900K] should be a real beast?

I was thinking the same thing. Could it be that a 10nm shrink of 8-core 9th-gen will replace Ice Lake-S for mainstream desktop? As far as I know, there weren't big architectural changes planned for Ice Lake (core count increase and AVX innovations mainly?). Seems to me the more substantial architectural innovation was originally planned for the subsequent microarchitecture generations (Sapphire Rapids, etc.). The Intel roadmap currently looks very foggy and uncertain, though.

What are your expectations? A final 14nm refresh next year, then a 10nm desktop introduction in 2020?

I do wonder if Intel has focused all of their efforts on the Rapids; and whether the future Lakes is going to be minimal updates

That makes sense. They now have a better feel for the competition to plan a competitive response. Meanwhile minor refinements and a shrink of Skylake/CFL will have to do, I guess.

 

[epsilon84]

Lisa Su's success as CEO doesn't hinge on Zen 2 beating a 9900K at gaming, not at all. I even doubt if Intel will be able to beat itself with a first gen Icelake chip, to be perfectly honest. We might be looking at a side grade but with better power consumption. We will see.

I don't doubt Zen 2 can match or surpass a 9900K in overall performance or price/performance or performance/watt, after all we are talking about a 7nm CPU against a 14nm one. What I do doubt is your unwavering belief that Zen 2 will beat a 9900K at gaming. AMD is coming from further behind in gaming than applications, which is because gaming is still largely ST performance dependant. It's a big gulf to overcome, that ST performance deficit.