Another, another Zen 2 Leak

[Zeeus]

Even if the CPU like previously leaked hits 5ghz, IT DEFS ISNT HITTING 5.8 i can almost bet my life on that thats how sure i am of that. But even if they manage a 5ghz boost, its literally RIP intel. Youll have better IPC than intel so faster gaming performance and with all the extra cores with AMD you already have better performance than intel in productivity so literally as they say "git fukd". I will be switching to zen 2 from my 8600k. 

[Sychic]

guys, it's not that hard

 

AMD will just include a DIY LN2 setup

[ravenshrike]

17 minutes ago, Zeeus said:

Youll have better IPC than intel so faster gaming performance

This should actually be the really interesting part. Given the leaked BIOS settings you're not going to see a big increase in IF speed, if any. So even with any IPC increases I'm not sure you'll see that large an increase in top FPS rates. However in theory minimum rates should see a rather large boost. So if clock speeds reach parity what you might end up with is AMD's minimum frame rates being significantly superior across the board while still lagging on the top end.

[Brooksie359]

48 minutes ago, Zeeus said:

Even if the CPU like previously leaked hits 5ghz, IT DEFS ISNT HITTING 5.8 i can almost bet my life on that thats how sure i am of that. But even if they manage a 5ghz boost, its literally RIP intel. Youll have better IPC than intel so faster gaming performance and with all the extra cores with AMD you already have better performance than intel in productivity so literally as they say "git fukd". I will be switching to zen 2 from my 8600k. 

The fact is that we have no idea how zen 2 would stack up against Intel of they manage to hit 5ghz. Programs act funny sometimes and ipc isn't the same across all applications so we womt know if they have better ipc in gaming scenarios. I would say it's better to wait for proper benchmarks rather than make claims that may or may not be true. 

[GoldenLag]

11 minutes ago, ravenshrike said:

This should actually be the really interesting part. Given the leaked BIOS settings you're not going to see a big increase in IF speed, if any. So even with any IPC increases I'm not sure you'll see that large an increase in top FPS rates. However in theory minimum rates should see a rather large boost. So if clock speeds reach parity what you might end up with is AMD's minimum frame rates being significantly superior across the board while still lagging on the top end.

https://www.notebookcheck.net/AMD-Ryzen-3000-s-MCM-design-enables-optimal-utilization-of-the-7-nm-process.414980.0.html

 

Do you mean latency or?

 

Because this says twice the bandwidth and AMD has stated lower overall latency.

[cj09beira]

4 hours ago, Stefan Payne said:

People rumored that they gave up the CCX per die -> 1 CCX = 1 Chiplet.

ccx configuration per die is the exact same, its been confirmed multiple times, and it is not a problem

19 minutes ago, ravenshrike said:

This should actually be the really interesting part. Given the leaked BIOS settings you're not going to see a big increase in IF speed, if any. So even with any IPC increases I'm not sure you'll see that large an increase in top FPS rates. However in theory minimum rates should see a rather large boost. So if clock speeds reach parity what you might end up with is AMD's minimum frame rates being significantly superior across the board while still lagging on the top end.

on zen 1 IF is not a limitation, latency and bandwidth are, if IF was the limitation ryzen would scale better with frequency than it does with latency, which is not the case, it scales just as well or better with latency

[CarlBar]

20 minutes ago, cj09beira said:

ccx configuration per die is the exact same, its been confirmed multiple times, and it is not a problem

on zen 1 IF is not a limitation, latency and bandwidth are, if IF was the limitation ryzen would scale better with frequency than it does with latency, which is not the case, it scales just as well or better with latency

 

Absolute minimum latency is at least partly tied to frequency.

[Zeeus]

59 minutes ago, ravenshrike said:

This should actually be the really interesting part. Given the leaked BIOS settings you're not going to see a big increase in IF speed, if any. So even with any IPC increases I'm not sure you'll see that large an increase in top FPS rates. However in theory minimum rates should see a rather large boost. So if clock speeds reach parity what you might end up with is AMD's minimum frame rates being significantly superior across the board while still lagging on the top end.

 

49 minutes ago, Brooksie359 said:

The fact is that we have no idea how zen 2 would stack up against Intel of they manage to hit 5ghz.

I would say it's better to wait for proper benchmarks rather than make claims that may or may not be true. 

Both of you are saying you dont know the speed... well i think its quite safe to assume they will be in the high 4 ghz well beyond 4.5 mark. Isnt it already confirmed with the zen refresh that the IPC of ryzen is better than intel counter parts? So with similar clocks to intel zen should be more powerful that much i believe is a fact, the only thing ryzen was lacking was clock speed. 

 

So lets assume a normal increase from the node shrink of 20%... last zen was at 4.3ghz so 20% increase takes that to 5.15ghz... thats pretty simple, lets assume worst case 15% clock increase would take it to 4.945ghz. Dudes its impossible (pretty unlikely) for it to get less than 15% improvement in clock speed with that huge of a node shrink, so saying 5ghz isnt CRAZY by any means. Even if they get 4.8ghz worst possible case, based on their IPC they should still outperform intel. 

Quote

Programs act funny sometimes and ipc isn't the same across all applications so we womt know if they have better ipc in gaming scenarios.

HUH? games run pretty well already across the board on all the ryzen cpus... i dont see how thats of even a remote concern. The only thing the games want is higher clocks. 

 

[Sychic]

1 minute ago, Zeeus said:

 

HUH? games run pretty well already across the board on all the ryzen cpus... i dont see how thats of even a remote concern. The only thing the games want is higher clocks. 

 

I think he was saying that the games can behave differently between AMD and Intel and that it might affect how the IPC in that program is

[Taf the Ghost]

3 hours ago, ravenshrike said:

This should actually be the really interesting part. Given the leaked BIOS settings you're not going to see a big increase in IF speed, if any. So even with any IPC increases I'm not sure you'll see that large an increase in top FPS rates. However in theory minimum rates should see a rather large boost. So if clock speeds reach parity what you might end up with is AMD's minimum frame rates being significantly superior across the board while still lagging on the top end.

Some of the first pass BIOS information wasn't quite right. The big thing is IF speed won't be tied to Memory speed completely, along with what appears to be some solutions to IF speed limitations. (The memory problems appear to more have been about the IF, which was tied to memory clocks, than the memory.) In theory, we should be able to clock the heck out of the IMC in the Zen2, which is really the space where there was problems.

 

The actual difference in gaming isn't IPC but optimization and cache speed + memory latency. The extra bit of clocks won't hurt, but they don't actually help from more processing. (Gaming almost never actually saturates a core for processing, but it saturates the I/O with the core.)

[Brooksie359]

3 hours ago, Zeeus said:

 

Both of you are saying you dont know the speed... well i think its quite safe to assume they will be in the high 4 ghz well beyond 4.5 mark. Isnt it already confirmed with the zen refresh that the IPC of ryzen is better than intel counter parts? So with similar clocks to intel zen should be more powerful that much i believe is a fact, the only thing ryzen was lacking was clock speed. 

 

So lets assume a normal increase from the node shrink of 20%... last zen was at 4.3ghz so 20% increase takes that to 5.15ghz... thats pretty simple, lets assume worst case 15% clock increase would take it to 4.945ghz. Dudes its impossible (pretty unlikely) for it to get less than 15% improvement in clock speed with that huge of a node shrink, so saying 5ghz isnt CRAZY by any means. Even if they get 4.8ghz worst possible case, based on their IPC they should still outperform intel. 

HUH? games run pretty well already across the board on all the ryzen cpus... i dont see how thats of even a remote concern. The only thing the games want is higher clocks. 

 

Ipc is application specific. There is no way for anyone to know how zen 2 ipc stacks up against Intel currently. Currently zen+ ipc compared to Intel is a mixed bag depending on the application. I think you shouldnt get all excited thinking that these new cpus are for sure going to beat Intel's current cpus in gaming. Its entirely possible that is the case. Its also entirely possible it isn't the case. I think it is dangerous to make claims like there is no way zen 2 doesn't beat Intel in gaming. I really do hope that zen 2 beats Intel in gaming but I am not going to expect it to because it's unfair to AMD to expect something like that. I would rather be pleasantly surprised and proud of AMD for taking the performance crown then be expecting that to be the case. If they dont take it I wont be disappointed because I wasn't expecting it. 

[CarlBar]

4 hours ago, Brooksie359 said:

Ipc is application specific. There is no way for anyone to know how zen 2 ipc stacks up against Intel currently. Currently zen+ ipc compared to Intel is a mixed bag depending on the application. I think you shouldnt get all excited thinking that these new cpus are for sure going to beat Intel's current cpus in gaming. Its entirely possible that is the case. Its also entirely possible it isn't the case. I think it is dangerous to make claims like there is no way zen 2 doesn't beat Intel in gaming. I really do hope that zen 2 beats Intel in gaming but I am not going to expect it to because it's unfair to AMD to expect something like that. I would rather be pleasantly surprised and proud of AMD for taking the performance crown then be expecting that to be the case. If they dont take it I wont be disappointed because I wasn't expecting it. 

 

Then you haven;t been paying attention. We've got a set of CPU's right now with a modest deficit compared to intel. AMD has then gone and raised IPC, Raised CLock speed, and significantly lowered IF link latency, (the latter being the real reason for the gap), AMD would have to mess up pretty hard to not at least achieve parity and they absolutly have potential to completely knock it out of the park.

 

In fact i've been doing some research into Infinity Fabric vs Intels Ring Bus and it's really highlighted both how bad the current situation is an how much of an upgrade Zen2 should be. Zen1/Zen+ actually has some pretty nasty issues with it's IF links when you start thinking what dataflow might look like under burst conditions and doing comparisons to Intel. Zen2 so massively ups the bandwidth and cuts the minimum latency that it should have a significant positive effect on many of the previous generations issues.

[mr moose]

5 minutes ago, CarlBar said:

 

Then you haven;t been paying attention. We've got a set of CPU's right now with a modest deficit compared to intel. AMD has then gone and raised IPC, Raised CLock speed, and significantly lowered IF link latency, (the latter being the real reason for the gap), AMD would have to mess up pretty hard to not at least achieve parity and they absolutly have potential to completely knock it out of the park.

 

In fact i've been doing some research into Infinity Fabric vs Intels Ring Bus and it's really highlighted both how bad the current situation is an how much of an upgrade Zen2 should be. Zen1/Zen+ actually has some pretty nasty issues with it's IF links when you start thinking what dataflow might look like under burst conditions and doing comparisons to Intel. Zen2 so massively ups the bandwidth and cuts the minimum latency that it should have a significant positive effect on many of the previous generations issues.

 

The only thing your lacking here is actual real world tests to show these things matter when comparing to Intel. 

 

The issue with most of these threads is that we don't have anything real to base said opinions on, it's all speculation until we have testing.  Gen2 might be the best thing since the Athlon 64 fx53. or might fall right behind Intel's next release.  Either way arguing projections based on assumptions and half released information isn't exactly going to prove anything (as everybody seems to be keen to see it succeed).

[Brooksie359]

32 minutes ago, mr moose said:

 

The only thing your lacking here is actual real world tests to show these things matter when comparing to Intel. 

 

The issue with most of these threads is that we don't have anything real to base said opinions on, it's all speculation until we have testing.  Gen2 might be the best thing since the Athlon 64 fx53. or might fall right behind Intel's next release.  Either way arguing projections based on assumptions and half released information isn't exactly going to prove anything (as everybody seems to be keen to see it succeed).

I'm glad someone said what I was trying to say in a more clear manner. I would love for AMD to knock it out of the park but we should wait until its released before we start rejoicing. 

[CarlBar]

2 hours ago, mr moose said:

 

The only thing your lacking here is actual real world tests to show these things matter when comparing to Intel. 

 

The issue with most of these threads is that we don't have anything real to base said opinions on, it's all speculation until we have testing.  Gen2 might be the best thing since the Athlon 64 fx53. or might fall right behind Intel's next release.  Either way arguing projections based on assumptions and half released information isn't exactly going to prove anything (as everybody seems to be keen to see it succeed).

 

Actually we've had some cache latency tests surface and whilst taking too many detailed comparisons would be premature a few other bits and piece of info have surfaced both officially and unofficially that are positive.

 

I agree going too far would be a mistake. But AMD would literally have to see near zero improvement to not stomp all over intel. the odd of that are not very high.