Is a 5Ghz Ryzen Cpu coming?

DuronBurgerMan said:
Yes, many modern games use more than 4 cores... kind of. There is a big jump from 4c/4t CPUs to 4c/8t or 6c/6t CPUs with same OC speed and uarch. There's a smaller jump from that to 6c/12t or 8c/8t. From that to 8c/16t, it's almost nothing (indeed, sometimes a regression, due to SMT overhead).

Now, that being said, a few years ago, we were struggling to see significant gains from > 4c/8t, and in some cases, 4c/4t. So in the last couple of years since Ryzen launched (and since Intel likewise started scaling core count), we have seen progress on this front. However, as some other folks pointed out, there are diminishing returns. Do I think we're stalled now? No, not really. But the gains to be had in gaming from core count scaling are not infinite. And they aren't coming quickly. And likely they will be tied to new game features.

Anyway, evidence in favor of my claim is below (from GN). You can see that 4c/4t CPUs are essentially worthless. You can also see that the highly OC'd 7700k is well behind its similarly-OCed 8700k, 9700k, and 9900k brothers, albeit still quite good (and better than the Zen+ line-up by a hair).

https://www.gamersnexus.net/images/media/2018/cpus/9700k/intel-i7-9700k-aco-1080p.png
Click to expand...

I wouldn't call 4C/4T worthless, for the vast majority of users out there, it is still pretty usable in games, a 7600K is still pretty decent for running vast majority of games, heck even R1200 is still pretty usable despite its low clock speed in most games.
 
Master_Pain said:
It's showing that, *gasp*, games use more than 4 cores. I upgraded back in the day from a 3770k to a 3930k, then to a 4930k, just for this game, to have more cores and then more cores with a higher clock.
Click to expand...

Is it? It shows no improved performance.

I remember looking at old single threaded games running on a quad core, showing the load spread among the cores even though it is single threaded. The scheduler is just moving the single thread between the cores.
 
Smoked Brisket said:
I just ordered a 3700x, a 570 mobo, and a 5700 xt. Anyone that makes a statement like the one above doesn't deserve to be listened to. I am purely a gamer, as I believe most builders are, even though they might be multitaskers. 9900k is a legit buy for that purpose, and the platform is not that much more, its comparable at this point in terms of cost. I was very on the fence and simply went with what I found more interesting and exciting at this time. I am not a moron, at least for that.
Click to expand...
I agree with you, you cannot go wrong buying 9900K is it is still relevant performance wise and being a mature platform, you don't have to deal with the early headaches of X570 motherboard.

BTW good choice on 3700X and enjoy your build.
 
Smoked Brisket said:
I just ordered a 3700x, a 570 mobo, and a 5700 xt. Anyone that makes a statement like the one above doesn't deserve to be listened to. I am purely a gamer, as I believe most builders are, even though they might be multitaskers. 9900k is a legit buy for that purpose, and the platform is not that much more, its comparable at this point in terms of cost. I was very on the fence and simply went with what I found more interesting and exciting at this time. I am not a moron, at least for that.
Click to expand...

Very on point. I like the way things have gone in the CPU market of late. At this point, you really can't go wrong with either brand. There are tradeoffs, of course, between one and the other, but both lines are excellent.
 
Smoked Brisket said:
I just ordered a 3700x, a 570 mobo, and a 5700 xt. Anyone that makes a statement like the one above doesn't deserve to be listened to. I am purely a gamer, as I believe most builders are, even though they might be multitaskers. 9900k is a legit buy for that purpose, and the platform is not that much more, its comparable at this point in terms of cost. I was very on the fence and simply went with what I found more interesting and exciting at this time. I am not a moron, at least for that.
Click to expand...

There are reasons to still buy Intel. Intel still has some advantage however in general advantages such as AVX / Linux stuff / specific development frameworks and a small handful of other niche things don't apply to average PC users and for sure don't apply to people that are strictly gamers.

If people are seriously arguing that a win of 1-2% at 1080p resolution means Intel wins gaming... then yes they are morons.

Where I live a 9900k + mobo will run you $200-300 (canadian) more then a 3700x + mobo. I don't know if your a pure gamer if paying 20-25% more for 1-2% at 1080p and 1080p alone makes a lick of sense. IMO people arguing that it does are yes morons.

Is it harsh... ya, yes I am being harsh. I guess I was just sick of listening to the AMD hate in every thread on every forum from Intel fan boys arguing... who cares about all the productivity blow outs. Intel still wins games. It's just such a completely stupid argument when Intels 1080p win is so low its almost less then 1% in most cases... and at 1440 and up FPS will be as good as the video card you have slotted. In your case you could have went 9900k instead and I guess only paid a hundred or so more if you where willing to buy a 5700 instead of a 5700 xt. To hit the same spend you would have had to give up even more. No doubt 3700x + 5700xt beats 9900k + 5700... and the Intel option would still be a good bit more $. Where I am $200-300 more for within the margin of error difference in CPU performance is pretty silly unless you have unlimited funds. (or a legit other reason to buy Intel... sure Photoshop still favors Intel, and there are other cases where Intel is still favored games just aren't a legit argument)
 
ChadD said:
There are reasons to still buy Intel. Intel still has some advantage however in general advantages such as AVX / Linux stuff / specific development frameworks and a small handful of other niche things don't apply to average PC users and for sure don't apply to people that are strictly gamers.

If people are seriously arguing that a win of 1-2% at 1080p resolution means Intel wins gaming... then yes they are morons.

Where I live a 9900k + mobo will run you $200-300 (canadian) more then a 3700x + mobo. I don't know if your a pure gamer if paying 20-25% more for 1-2% at 1080p and 1080p alone makes a lick of sense. IMO people arguing that it does are yes morons.

Is it harsh... ya, yes I am being harsh. I guess I was just sick of listening to the AMD hate in every thread on every forum from Intel fan boys arguing... who cares about all the productivity blow outs. Intel still wins games. It's just such a completely stupid argument when Intels 1080p win is so low its almost less then 1% in most cases... and at 1440 and up FPS will be as good as the video card you have slotted. In your case you could have went 9900k instead and I guess only paid a hundred or so more if you where willing to buy a 5700 instead of a 5700 xt. To hit the same spend you would have had to give up even more. No doubt 3700x + 5700xt beats 9900k + 5700... and the Intel option would still be a good bit more $. Where I am $200-300 more for within the margin of error difference in CPU performance is pretty silly unless you have unlimited funds. (or a legit other reason to buy Intel... sure Photoshop still favors Intel, and there are other cases where Intel is still favored games just aren't a legit argument)
Click to expand...

I just want to address a couple points:

1. The performance difference in games is ~6% at 1080p (with high end GPU), not 1%. This is against the 3900X (although it also applies to the 3800X, which is close-ish in game performance). So not margin of error.
2. AMD actually wins in Photoshop, albeit it not by much. A couple percentage points at most. It depends on the benchmark in question, but the Puget Systems benchmark is pretty thorough and grants the 3900X a victory. If you break it down further, you will find certain Photoshop tasks that favor one or the other.
3. The economic argument in favor of the 3700X holds true for a lot of use cases, but 100% pure gaming... if your budget is stuck, a 9700k is probably a better buy (or maybe roughly the same) than a 3700X. But that's a much closer argument, really.
 
The best part about all of this is that we are finally arguing single digit percentage differences between AMD and Intel for the first time in a long time. If you are building today and have the budget for 9900k or 3900x systems you actually shouldn't be disappointed by either. Great stuff going on right now.
 
DuronBurgerMan said:
I just want to address a couple points:

1. The performance difference in games is ~6% at 1080p (with high end GPU), not 1%. This is against the 3900X (although it also applies to the 3800X, which is close-ish in game performance). So not margin of error.
2. AMD actually wins in Photoshop, albeit it not by much. A couple percentage points at most. It depends on the benchmark in question, but the Puget Systems benchmark is pretty thorough and grants the 3900X a victory. If you break it down further, you will find certain Photoshop tasks that favor one or the other.
3. The economic argument in favor of the 3700X holds true for a lot of use cases, but 100% pure gaming... if your budget is stuck, a 9700k is probably a better buy (or maybe roughly the same) than a 3700X. But that's a much closer argument, really.
Click to expand...

1... 6% depends on the title, who did the test, and what GPU exactly was used. 6% is an absolute best case scenario for Intel. In most cases it is margin of error... and again afer 1080p I haven't seen anyone claiming anything more then a couple % points. Still point taken... yes perhaps a couple titles still favor intel slightly. Although for the first time ever there are titles that actually favor AMD so I consider it a wash. They couldn't be more then equal if Intel and AMD worked together on it. ;)
2... fair I haven't looked that close at photoshop benches. I think with stuff like photoshop and high end software like Houdine it often comes down to the plugins / external renders being used. I know there are some cases of specific things favoring Intel still. No knock on Intel their software guys have always done a good job of getting things heavily optimized. Again that stuff just doesn't apply to gamers.
3... I guess you can argue 3700x vs 9700k... I just don't see how Intel wins that one. Pricing is close with Intel only being slighly more expensive. You give up SMT for a bit more mhz.... and real world benches they are damn near identical. Your right I guess you can argue the Intel chip makes sense if you really are a PURE gamer and would prefer hyper threading off. Still at stock speeds the 3700x will have faster single and dual core speeds. If you are OCing the 9700k may take a small win in single core... but again zero SMT/HT support. 3700x over 9700k... slightly (1-6%) lower single core for a healthy 20+% gain in multi core. The Intel option doesn't seem that attractive to me if you don't have non game reasons to go Intel. Vs the more apples to apples 9900k your going to spend a lot more to go Intel for the same single digit gains in single core and a slight less loss in multi core.
 
The 9900k is 15 dollars cheaper than the 3900x, the hero 11 is 100 dollars cheaper than the hero 3. Again, I just ordered a full AMD build, but lets get the facts straight. We can all cherry pick benchmarks and claim that in our respective towns that AMD apples to apples is still cheaper(you cannot compare the 9900k to the 3700k), but the 9900k is right now the better gaming cpu, and cheaper than the 3900x. I am not saying that it is overall better, but for mine, and many others use it is a legit choice. There is no AMD hate in forums such as these. Quite the opposite to the point where many cannot look at this competition objectively. All that does is diminish AMD's comeback, which has been quite spectacular.
 
andrewaggb said:
I have an 8700k with 3 monitors I use for work (at home). Most of the time I just leave most apps open and games still run fine. I do have a separate gaming pc but the convenience factor of running games on my work pc (I work from home) and not closing everything is pretty high. I'm not super competitive but I certainly enjoy that I don't have to shut everything down. Years ago (with windows xp) I used to turn off so much stuff that task manager reported 14 processes before I'd play a game. Right now my pc says 351 processes and 16.4gb of memory are in use, and I'd be totally fine to launch a game and have it run well. That's a pretty big quality of life improvement.
Click to expand...

Isn't it great that we can do that?
The cost to entry to great multitasking is getting lower and lower. Progress is great, can't wait till the 3950X and intel's competitor drops so that I can start looking at upgrades.... I really want that 3950X, but it will cost so much...
 
ChadD said:
1... 6% depends on the title, who did the test, and what GPU exactly was used. 6% is an absolute best case scenario for Intel. In most cases it is margin of error... and again afer 1080p I haven't seen anyone claiming anything more then a couple % points. Still point taken... yes perhaps a couple titles still favor intel slightly. Although for the first time ever there are titles that actually favor AMD so I consider it a wash. They couldn't be more then equal if Intel and AMD worked together on it. ;)
Click to expand...

6% is the average across many games. Obviously, AMD does win some individual benches. But 6% isn't a best case scenario, it is the mean. Hardware Unboxed did a 36 game test - look that up if you like. 6% in 1080p is not a particularly big deal. But the difference exists.

3... I guess you can argue 3700x vs 9700k... I just don't see how Intel wins that one. Pricing is close with Intel only being slighly more expensive. You give up SMT for a bit more mhz.... and real world benches they are damn near identical. Your right I guess you can argue the Intel chip makes sense if you really are a PURE gamer and would prefer hyper threading off. Still at stock speeds the 3700x will have faster single and dual core speeds. If you are OCing the 9700k may take a small win in single core... but again zero SMT/HT support. 3700x over 9700k... slightly (1-6%) lower single core for a healthy 20+% gain in multi core. The Intel option doesn't seem that attractive to me if you don't have non game reasons to go Intel. Vs the more apples to apples 9900k your going to spend a lot more to go Intel for the same single digit gains in single core and a slight less loss in multi core.
Click to expand...

9700k vs 3700X is a harder sell. I would highly recommend the 3700X for anybody except a pure, >95% gamer, in that price bracket. The extra threads can make a large difference in any workload that can use them, meanwhile the gain for the 9700k in gaming is small. But it exists, and there are users that are >95% gamers.

Consider that in terms of IPC, Zen has always had a higher throughput speed than latency-sensitive speed. Zen 2 partially obfuscates this by adding boatloads of extra cache. Zen 1, for instance, was ~Haswell-ish (or maybe even Ivy Bridge-ish) in terms of latency-sensitive performance, but was Broadwell-ish in terms of throughput per core. And then, of course, brought 2x the cores to the table. Zen 2 is broadly Skylake-ish in IPC, albeit it slightly behind, in latency-sensitive workloads, though Intel's clockspeed advantage means Intel wins here (but not by much). In throughput, Zen 2 has superior IPC to Skylake. This is why in Cinebench R20, the 3900X usually beats the 9900k even in single thread output, despite a large clockspeed deficit. This explains Zen 2's bifurcated performance. Workloads like Photoshop are a mix of latency-sensitive and throughput tasks, and so Zen 2 can pull wins and/or ties, but gaming is predominantly latency-sensitive. In throughput tasks, the 3700X sometimes beats the 9900k!

So the performance matrix works like this:

1. Throughput tasks: Zen 2 wins, often even at the same core count. But almost always at the same price point (more cores/threads per dollar). (Exception: some AVX512-friendly workloads - but even there, more cores often wins the day).
2. Throughput/Latency mix: Zen 2 and CFL are competitive. One or the other could win, at the same core count. In the same price bracket, with more cores/threads per dollar, AMD wins more often, but by no means all the time. And the margins tend to be slim.
3. Latency tasks: CFL wins at the same core/thread count, and usually wins even when more cores are brought to the table. (Exception: lower-end non-k parts where Intel does not maintain a clockspeed advantage).
 
DuronBurgerMan said:
6% is the average across many games. Obviously, AMD does win some individual benches. But 6% isn't a best case scenario, it is the mean. Hardware Unboxed did a 36 game test - look that up if you like. 6% in 1080p is not a particularly big deal. But the difference exists.



9700k vs 3700X is a harder sell. I would highly recommend the 3700X for anybody except a pure, >95% gamer, in that price bracket. The extra threads can make a large difference in any workload that can use them, meanwhile the gain for the 9700k in gaming is small. But it exists, and there are users that are >95% gamers.

Consider that in terms of IPC, Zen has always had a higher throughput speed than latency-sensitive speed. Zen 2 partially obfuscates this by adding boatloads of extra cache. Zen 1, for instance, was ~Haswell-ish (or maybe even Ivy Bridge-ish) in terms of latency-sensitive performance, but was Broadwell-ish in terms of throughput per core. And then, of course, brought 2x the cores to the table. Zen 2 is broadly Skylake-ish in IPC, albeit it slightly behind, in latency-sensitive workloads, though Intel's clockspeed advantage means Intel wins here (but not by much). In throughput, Zen 2 has superior IPC to Skylake. This is why in Cinebench R20, the 3900X usually beats the 9900k even in single thread output, despite a large clockspeed deficit. This explains Zen 2's bifurcated performance. Workloads like Photoshop are a mix of latency-sensitive and throughput tasks, and so Zen 2 can pull wins and/or ties, but gaming is predominantly latency-sensitive. In throughput tasks, the 3700X sometimes beats the 9900k!

So the performance matrix works like this:

1. Throughput tasks: Zen 2 wins, often even at the same core count. But almost always at the same price point (more cores/threads per dollar). (Exception: some AVX512-friendly workloads - but even there, more cores often wins the day).
2. Throughput/Latency mix: Zen 2 and CFL are competitive. One or the other could win, at the same core count. In the same price bracket, with more cores/threads per dollar, AMD wins more often, but by no means all the time. And the margins tend to be slim.
3. Latency tasks: CFL wins at the same core/thread count, and usually wins even when more cores are brought to the table. (Exception: lower-end non-k parts where Intel does not maintain a clockspeed advantage).
Click to expand...

well said
 
Master_Pain said:
It's showing that, *gasp*, games use more than 4 cores. I upgraded back in the day from a 3770k to a 3930k, then to a 4930k, just for this game, to have more cores and then more cores with a higher clock.
Click to expand...

I didn't argue against this. I argued- paraphrasing without going back to quote myself- that your average consumer doesn't really need more than four cores, and that is absolutely true.

For gaming, I'd argue for at least six cores with SMT, the higher the clockspeed and IPC, the better. That means that even if AMD could produce a 5GHz Ryzen 3000 (they can't and won't), they'd still be behind current Intel parts for gaming.

Now, how much behind along with cost is another discussion, and where I've addressed that elsewhere in the forum, you'll find that I've recommended AMD where the user in question is not looking to maximize gaming performance.
 
IdiotInCharge said:
I didn't argue against this. I argued- paraphrasing without going back to quote myself- that your average consumer doesn't really need more than four cores, and that is absolutely true.

For gaming, I'd argue for at least six cores with SMT, the higher the clockspeed and IPC, the better. That means that even if AMD could produce a 5GHz Ryzen 3000 (they can't and won't), they'd still be behind current Intel parts for gaming.

Now, how much behind along with cost is another discussion, and where I've addressed that elsewhere in the forum, you'll find that I've recommended AMD where the user in question is not looking to maximize gaming performance.
Click to expand...


If it takes an AC OC to 5Ghz+ for Intel to eek out a 6-10% lead at 1080p, while dropping to low single digits at 1440+, against a 4.3Ghz 3700x, then how do you justify the claim that "AMD would still be behind for gaming" with a 5Ghz same core count SKUs?


They would be besting Intel @4.6-4.7Ghz, and at 5Ghz would most likely give them a clear double digits lead.


I just don't understand your logic...Maybe you failed to be clear in what you are really saying but with the way things stand now, 700Mhz in extra headroom with AMD's advantsge in IPC would give them a clear win.
 
Zen 2's IPC advantage is in throughput tasks, not latency-sensitive tasks (like gaming). I think it was also Hardware Unboxed that locked the 3800X to 4GHz, locked the 2700X to 4GHz, and the 9900k to 4GHz, and then ran a series of gaming benchmarks. Zen 2 was far ahead of Zen+, but still behind the 9900k - albeit the percentage was even smaller than currently exists in the wild, where Intel maintains a clockspeed advantage.

In other words, if you had a 5GHz 8 core Zen 2 offering (you won't, the process won't take it that far w/o LN2 this generation), it would be about maybe 3% slower than the 9900k, instead of 6% slower.

OTOH, conduct the same locked clockspeed test in a throughput task like rendering with equal core count CPUs... and Zen 2 will win, and even does now often times despite the clockspeed disadvantage.

It's the nature of the uarch's to be this way. AMD made a tradeoff with Zen - both the original incarnation, and the new one. They chose a modular approach with a fast interconnect, instead of a monolithic setup with a ring bus. That imposed a cost in latency that must exist (though AMD has worked hard to mitigate it as much as possible, and done well at it), and even Zen 2's massive amount of cache can't entirely eliminate it. The benefit is incredibly easy/cheap core scaling with CCXs (or integrated GPUs for the APUs) on each die, and multiple dies. With the separation of the I/O, theoretically AMD can scale this as much as they want. This latency cost does not affect throughput performance much, and so there Zen derivatives really show their muscle.

IMHO, AMD made the right call given their R&D budget limitations, the current market conditions, and Intel's stalled position re: 10nm process. They've done very well. But they do not win all the things. And there's no reason to spin that way it like some folks are doing. For mixed-use (gaming, workstation, etc... all-in-one), go AMD! For workstation/content creation/coding/etc... Go AMD! For max gaming FPS, go Intel!
 
Revdarian said:
https://www.eteknix.com/amd-details-longterm-zen-cpu-roadmap/
Here's the current roadmaps, as you can see on the 7nm+ they only have plans for Zen3 at the moment. Zen2 is what we have and what we have is Zen2, temper your expectations of anything extra being thrown at it.
Click to expand...

well according to that roadmap, Zen3 basically is Zen2+. jus sayin.

Revdarian said:
Maybe, maybe not. Didn't they say that they were planning to move to a new socket too?
Click to expand...

hopefully not but if it means getting away from a fan cooled chipset. guess you gotta do what you gotta do.

will be interesting to see if intel also needs a fan when they move to PCIE 4.0
 
Snowdog said:
Most people will just be doing:
Basic productivity ( Office SW, Tax SW). Won't tax a 4 core.
Media consumption. Won't tax a 4 core.
Web/Internet usage. Won't tax a 4 core.
Casual gaming. Won't tax a 4 core.
Click to expand...

dude what are you talking about? if you have enough tabs open and loading pages today you can use more than 50% on 8 core. there are plenty of games that benefit from more than 4 cores plus you have background tasks that may need processor cycles. besides all the new gaming consoles will have 8 core proc's. AND media encoding.. No one likes to wait for encoding to finish, i don't care how long it takes, so the more cores the better. What about streamers???!

you do realize you're posting in a [H]ard forum, right?
 
ccityinstaller said:
If it takes an AC OC to 5Ghz+ for Intel to eek out a 6-10% lead at 1080p, while dropping to low single digits at 1440+, against a 4.3Ghz 3700x, then how do you justify the claim that "AMD would still be behind for gaming" with a 5Ghz same core count SKUs?


They would be besting Intel @4.6-4.7Ghz, and at 5Ghz would most likely give them a clear double digits lead.


I just don't understand your logic...Maybe you failed to be clear in what you are really saying but with the way things stand now, 700Mhz in extra headroom with AMD's advantsge in IPC would give them a clear win.
Click to expand...

It's not my logic.

See DuronBurgerMan's post above. Ryzen 3000 has been tested to have lower IPC at the same clockspeed in gaming. Worse, there are still frametime issues that affect the 0.1% lows, and those are the issues that you would feel.


Further: your position makes the basic fallacious assumption concerning the relationship of performance at particular resolutions. We use 'faster' and 'slower' for a reason: if a GPU twice as fast as a 2080Ti were available, and that's something that we can bet on happening eventually, 1080p performance on a 2080Ti would certainly be applicable to that future GPU at 1440p and even 4k.

Buying a CPU because it's 'fast enough at 1440p / 4k today' makes little sense if like most people that CPU would be kept through multiple GPU upgrades.
 
IdiotInCharge said:
It's not my logic.

See DuronBurgerMan's post above. Ryzen 3000 has been tested to have lower IPC at the same clockspeed in gaming. Worse, there are still frametime issues that affect the 0.1% lows, and those are the issues that you would feel.


Further: your position makes the basic fallacious assumption concerning the relationship of performance at particular resolutions. We use 'faster' and 'slower' for a reason: if a GPU twice as fast as a 2080Ti were available, and that's something that we can bet on happening eventually, 1080p performance on a 2080Ti would certainly be applicable to that future GPU at 1440p and even 4k.

Buying a CPU because it's 'fast enough at 1440p / 4k today' makes little sense if like most people that CPU would be kept through multiple GPU upgrades.
Click to expand...

This post resonates with me. My main rig is a 2700x and I am looking to replace a kid’s rig Ryzen 1400 with it.

I was looking at the 3950x. The thing is I get frametime spikes in VR with the 2700x, which VR is my main performance driver. I never got these with my 5960x which should be slightly slower.

The second issue is I want to get wireless VR for xmas. I have read about it not working with some Ryzen mobos for one reason or another.

I am now curious if Intel is launching anything before christmas. I am out of the loop but guessing no.
 
Dayaks said:
I am now curious if Intel is launching anything before christmas. I am out of the loop but guessing no.
Click to expand...

Rumors of a 10-core for 115x, but no solid details. If you need to keep frametimes down for VR though, and that's a provable problem with Ryzen, then a 9900K isn't a hard sell. I won't say it's an easy one with more cores available, but unless you're using those cores for something other than gaming, why not make what you do work?
 
IdiotInCharge said:
It's not my logic.

See DuronBurgerMan's post above. Ryzen 3000 has been tested to have lower IPC at the same clockspeed in gaming. Worse, there are still frametime issues that affect the 0.1% lows, and those are the issues that you would feel.


Further: your position makes the basic fallacious assumption concerning the relationship of performance at particular resolutions. We use 'faster' and 'slower' for a reason: if a GPU twice as fast as a 2080Ti were available, and that's something that we can bet on happening eventually, 1080p performance on a 2080Ti would certainly be applicable to that future GPU at 1440p and even 4k.

Buying a CPU because it's 'fast enough at 1440p / 4k today' makes little sense if like most people that CPU would be kept through multiple GPU upgrades.
Click to expand...


I never said that you would not seen a difference down the road once we get new GPUs, that you know, allow us to remove/raise the bar in the GPU being the bottleneck. This hold true for ANY CPU on the market, so I'm not sure exactly what you saying.


That in 5-9 years when a 2080ti is considered the 1060 6GB/rx 590 of today that the extra 6% Intel currently has will come in handy? Sure I agree with that 100%.

That does not nothing to acknowledge the fact that even in a latency sensitive workload, giving Zen2 a 5Ghz clock speed would put it on par or ahead.

I'm well aware that we aren't going to get there until possibly Zen3, if ever considering how wide the Zen2 uarch already is and that AMD may again attempt to widen it once more since that is one way to increase IPC if you do not have a ton of available clock speed headroom...W

will have to see how the 7+, and 6nm nodes fair for TSMC.

I wish I had still had a water chiller so we could see how a 4.6-4.7Ghz 3700x would fair with 3600c14-16 ram vs a 9900k @ 5Ghz. I do not miss that electric bill through!

DuronBurgerMan , thanks for clarifying what he was saying.
 
ccityinstaller said:
I never said that you would not seen a difference down the road once we get new GPUs, that you know, allow us to remove/raise the bar in the GPU being the bottleneck. This hold true for ANY CPU on the market, so I'm not sure exactly what you saying.
Click to expand...

I'm saying what needs to be said: you made an argument that holds true only if it isn't said.

ccityinstaller said:
That does not nothing to acknowledge the fact that even in a latency sensitive workload, giving Zen2 a 5Ghz clock speed would put it on par or ahead.
Click to expand...

And this is wrong: AMD made latency worse on Zen 2 / Ryzen 3000 by decoupling the memory controller. We're still finding issues, and we'll continue finding issues for years. It's a major architectural departure for desktop software, and in many ways, a step backward. Kudos to AMD for hiding it so well, but the impact of that decision hasn't been fully felt either, and just boosting clockspeeds does nothing for latency issues. In fact, higher clockspeeds are likely to make latency issues more apparent, because the CPU will not scale linearly as would be expected from an Intel CPU with a monolithic architecture that lacks wacky bus clocks to throw performance off.

ccityinstaller said:
will have to see how the 7+, and 6nm nodes fair for TSMC.
Click to expand...

Flip a coin. Intel messed up their 10nm node to be sure, but they also have a far better track record of executing new manufacturing processes with large, complex, high-clockspeed parts than TSMC or Samsung. Maybe AMD and TSMC pull off their next few releases- but I wouldn't dare count Intel out or bet on them to make another misstep. And don't take that as pro- or anti- whatever; that's just how AMD and Intel have performed.
 
I am considering the 9700K and 3700X for my next build and was leaning towards the 3700X. It seems the better Cpu in general if one plans to use it outside of gaming in heavily threaded workloads. Gaming performance is significantly improved over the previous generation, is close to Intel above 1080P and is certainly "good enough." That's even before you factor in other positives for the AMD processor like lower power consumption and a decent cooler included. Other reasons compelling me towards 3700X were the valid and widely expressed sentiments that Intel has been milking it's customers with very small generational IPC improvements, inflated pricing and low core counts for far too long. I very much want the underdog to succeed and may yet open my wallet to do my bit to help them achieve that goal and so in a way it pains me to say some very good points have been made in favor of Intel in this thread. Most importantly, but for me obfuscated by a warm glow of goodwill towards AMD, if your number 1 consideration for a new machine presently is gaming performance, the 9700K is probably the better buy. The decision is going to be harder than I thought.
 
one thing i do not get, the whole intell / amd gaming war, its great to see these benchmarks but on a gaming system high end or low, with the same GPU will ya really notice the difference? talking about somewhat comparable cpu's...9700/3700. is 5-10 fps or usually less than that, a dealbreaker when both are over 100fps? seems like its all kind of overblown, not like one only gets 30FPS and the other gets 100..i remember years ago 2 different company's arguing about who bedsheets were the best, one had like 800 threds per inch and the other had 850.....sorry, WGAF, they were both pretty dam good.
 
  • Like
Reactions: Auer
like this
Dayaks said:
This post resonates with me. My main rig is a 2700x and I am looking to replace a kid’s rig Ryzen 1400 with it.

I was looking at the 3950x. The thing is I get frametime spikes in VR with the 2700x, which VR is my main performance driver. I never got these with my 5960x which should be slightly slower.

The second issue is I want to get wireless VR for xmas. I have read about it not working with some Ryzen mobos for one reason or another.

I am now curious if Intel is launching anything before christmas. I am out of the loop but guessing no.
Click to expand...
I would grab a 3700x or 3800x and try that then look at a new platform if it's still an issue.


I love this justification for a 6% CPU difference in gaming making a difference in playability..... if in future that few percentage is an issue you will be upgrading the whole computer anyway. 6% is NOTHING, this is bonkers amounts of FUD.

IdiotInCharge said:
And this is wrong: AMD made latency worse on Zen 2 / Ryzen 3000 by decoupling the memory controller. We're still finding issues, and we'll continue finding issues for years. It's a major architectural departure for desktop software, and in many ways, a step backward. Kudos to AMD for hiding it so well, but the impact of that decision hasn't been fully felt either, and just boosting clockspeeds does nothing for latency issues. In fact, higher clockspeeds are likely to make latency issues more apparent, because the CPU will not scale linearly as would be expected from an Intel CPU with a monolithic architecture that lacks wacky bus clocks to throw performance off.
Click to expand...
WRONG.gif


No they didn't make all latency worse, mr blanket statement, are you related to Shintai? Only increase was memory latency which accompanied an increase in throughput and maximum rate.

upload_2019-8-11_22-33-52.png

source:plebbit
 
B770 said:
one thing i do not get, the whole intell / amd gaming war, its great to see these benchmarks but on a gaming system high end or low, with the same GPU will ya really notice the difference? talking about somewhat comparable cpu's...9700/3700. is 5-10 fps or usually less than that, a dealbreaker when both are over 100fps? seems like its all kind of overblown, not like one only gets 30FPS and the other gets 100..i remember years ago 2 different company's arguing about who bedsheets were the best, one had like 800 threds per inch and the other had 850.....sorry, WGAF, they were both pretty dam good.
Click to expand...
Something within 6% of playable framerates will also have about the same minimums - 5 years time if you relied on that tiny amount you would be replacing the whole setup, both would be practically the same experience. Meanwhile as more streaming/etc programs come out that run in background, the MT throughput of Zen2 will have it's own advantage.

AKA Intel is faultless and 6% difference on intel is actually 6000% faster in 10 years time in an AVX1064 game.
 
Ranger101 said:
valid and widely expressed sentiments that Intel has been milking it's customers with very small generational IPC improvements, inflated pricing and low core counts for far too long. I very much want the underdog to succeed and may yet open my wallet to do my bit to help them achieve that goal and so in a way it pains me to say some very good points have been made in favor of Intel in this thread. Most importantly, but for me obfuscated by a warm glow of goodwill towards AMD, if your number 1 consideration for a new machine presently is gaming performance, the 9700K is probably the better buy. The decision is going to be harder than I thought.
Click to expand...

I don't see it as that difficult- here's how I'd do it today (I own an 8700k and 9900k, both bought before AMD started getting their shit together): if I were buying a CPU for enthusiast purposes, that's gaming or other intensive work, I'd get no less than eight SMT cores.

That's the 9900K, 1700X+ / 2700X+ / 3700X+, and HEDT. Which way I go after that would entirely be a function of priorities. The 9900K is the fastest gaming CPU, but it's also a hard sell when you can get four more cores for about the same price.

So, for outright pushing-the-limits gaming, 9900K.

But just for general gaming? 3700X. General workstation? Hell, 2700X. Hard budget? 1700X. As available for older CPUs of course, they'll probably disappear.

And for general gaming + heavy workstation usage, 3900X or wait for 3950X.


[and with respect to the politics and feelings and so on- if you buy a less optimal alternative, you're just making a donation to a cold-blooded corporation that will happily take you for all you're worth given the opportunity; there are no morals here, it's just business]
 
N4CR said:
I would grab a 3700x or 3800x and try that then look at a new platform if it's still an issue.

I love this justification for a 6% CPU difference in gaming making a difference in playability..... if in future that few percentage is an issue you will be upgrading the whole computer anyway. 6% is NOTHING, this is bonkers amounts of FUD.

No they didn't make all latency worse, mr blanket statement, are you related to Shintai? Only increase was memory latency which accompanied an increase in throughput and maximum rate.

source:plebbit
Click to expand...

The FUD is claiming that a 6% average delta is representative of what a single end-user would experience. That would only be true if they played every game in the benchmark in equal proportion. Let's see how many people buy that argument.

And yes, AMD absolutely did make latency worse, literally by design. But of course you didn't link memory latency, because that wouldn't support your fallacious rebuttal ;)
 
N4CR said:
AKA Intel is faultless and 6% difference on intel is actually 6000% faster in 10 years time in an AVX1064 game.
Click to expand...

Do not put words in my mouth and make assumptions about what I would say. If you're that curious, you can ask specific questions instead of trolling.
 
You must log in or register to reply here.
Back
Top