AMD Zen 2 (Ryzen 3000) - *** NO COMPETITOR HINTING ***

[Deleted User 456458]

We don’t a lot of info to go on so we are back to speculating. However, the key here is that we will get a Ryzen 2 soon and expect leaks leading upto it.

Once CES settles down there will be more info coming out. The clockspeed of that ES is probably the biggest unknown variable we have at this time.

 

[beany_bot]

We don’t a lot of info to go on so we are back to speculating. However, the key here is that we will get a Ryzen 2 soon and expect leaks leading upto it.

Once CES settles down there will be more info coming out. The clockspeed of that ES is probably the biggest unknown variable we have at this time.

The only thing we have to go on is this

Ryzen3xxx was show beating 9900k 8c vs 8c.

Plenty good enough for me to ready my wallet for June.

 

[KentMan]

Quite simply AMD put up their chip, on a like for like basis, wouldn't surprise me if that AMD chip was clocked around 4.5ghz either. So they put their 8c 16t vs Intels 8c 16t chip, unfortunately in this instance for Intel, AMD's chip while being an ES and probably still not the final clocks, was matching it for performance at a lower power, if theres room left in that tank then AMD have beaten Intels high end with their own mid range, as AMD have an ace up their sleeve with another set of cores able to be bolted on.

The only question i have with that regard though is heat... we dont know how hot that 8c 16t AMD chip was, it was on an air cooler sure, but was that a Wraith Cooler? or was it like a Phanteks level of air cooler? we cannot tell, as we have no specifics. Im inclined to believe that the power draw was pointing at that AMD chip not being aggressively clocked and still matching a stock 9900k.

Now if we put 2 of those chips on a package the size of ryzen thats a lot of heat we need to dissipate, they could be clocked lower to keep thermals in check, we just dont know.

As someone above has said, with yesterday out of the way we should now start seeing more leaks coming through about actual specifics. If the X570 motherboards are being touted around to reviewers already this could be very good as well.

Also i read somewhere last night that PCIE4.0 may well be able to be pushed to First and 2nd Gen Zen motherboards? big if true.

 

[beany_bot]

Quite simply AMD put up their chip, on a like for like basis, wouldn't surprise me if that AMD chip was clocked around 4.5ghz either. So they put their 8c 16t vs Intels 8c 16t chip, unfortunately in this instance for Intel, AMD's chip while being an ES and probably still not the final clocks, was matching it for performance at a lower power, if theres room left in that tank then AMD have beaten Intels high end with their own mid range, as AMD have an ace up their sleeve with another set of cores able to be bolted on.

The only question i have with that regard though is heat... we dont know how hot that 8c 16t AMD chip was, it was on an air cooler sure, but was that a Wraith Cooler? or was it like a Phanteks level of air cooler? we cannot tell, as we have no specifics. Im inclined to believe that the power draw was pointing at that AMD chip not being aggressively clocked and still matching a stock 9900k.

Now if we put 2 of those chips on a package the size of ryzen thats a lot of heat we need to dissipate, they could be clocked lower to keep thermals in check, we just dont know.

As someone above has said, with yesterday out of the way we should now start seeing more leaks coming through about actual specifics. If the X570 motherboards are being touted around to reviewers already this could be very good as well.

Also i read somewhere last night that PCIE4.0 may well be able to be pushed to First and 2nd Gen Zen motherboards? big if true.

I think 7nm should help a lot with heat? Can't be any worse than the furnace that is the 9900k.

 

[Potatowithearsontheside]

So you reckon AMD have a near 20% IPC advantage over Intel now, then? Interesting.

Assuming Zen 2 isn't a regression from Zen+, we know the chip can't have been running at over 4.7 GHz. It was likely running somewhere between 4 and 4.5 GHz.

DG was likely being incredibly sarcastic. The naysayers are suffering from a touch of cognitive dissonance anyway. Either the clocks are incredibly high, or the IPC is significantly ahead, however both of these things have been labeled as impossibilities by Intel shills. The CB results show that one or the other must be true, and its easier for them to dismiss any gains by focusing on an extreme (even though that extreme implies that the other impossibility must be true as a result)...and amusingly they switch between those extremes dependent upon how they want to frame a response. Its laughable.

If their responses were more akin to "OK, so we know that CB 15 MT represents a best case scenario for Ryzen, and we also know that AMD's implementation of SMT is more efficient than Intel's, so we can see that AMD has achieved parity at multithread on a core to core comparison, but we can also say that it MAY NOT have achieved parity on ST performance as we'd expect to see MT ahead if there was parity on ST" then the shills wouldn't be getting so much heat.
It's true to say that we don't know the ST performance of Ryzen 3xxx, but it's dishonest to suggest that Intel is still ahead without further information. Same goes for AMD. We can speculate, primarily based around approximate clock speeds of that ES, but little more.

 

[Journey]

Now if we put 2 of those chips on a package the size of ryzen thats a lot of heat we need to dissipate, they could be clocked lower to keep thermals in check, we just dont know.

You need to consider that the actual chiplet with the CPU cores is not the only thing drawing power, but the the 14nm I/O as well. Given the way that the interconnect works, the power requirements for the I/O die shouldn't increase too much by adding a second chiplet with a further 8 cores, so you are not looking at 75w x2 but more like 75w + 20-30w, using the example shown.

If they do release a 3850X style product, with 16c/32t at 135w TDP there is reasonable certainty that this would only be done at higher clock speeds to prove a point, and make flagship part. I mean using the example shown by AMD yesterday the Intel 8c part was pulling 120w+ at 4.7GHz.

 

[beany_bot]

Any day now... "Awrite guys hows it going"...

 

[DragonQ]

DG was likely being incredibly sarcastic.

I know, I just feel like pointing out his nonsense sometimes.

The naysayers are suffering from a touch of cognitive dissonance anyway. Either the clocks are incredibly high, or the IPC is significantly ahead, however both of these things have been labeled as impossibilities by Intel shills. The CB results show that one or the other must be true, and its easier for them to dismiss any gains by focusing on an extreme (even though that extreme implies that the other impossibility must be true as a result)...and amusingly they switch between those extremes dependent upon how they want to frame a response. Its laughable

If their responses were more akin to "OK, so we know that CB 15 MT represents a best case scenario for Ryzen, and we also know that AMD's implementation of SMT is more efficient than Intel's, so we can see that AMD has achieved parity at multithread on a core to core comparison, but we can also say that it MAY NOT have achieved parity on ST performance as we'd expect to see MT ahead if there was parity on ST" then the shills wouldn't be getting so much heat.
It's true to say that we don't know the ST performance of Ryzen 3xxx, but it's dishonest to suggest that Intel is still ahead without further information. Same goes for AMD. We can speculate, primarily based around approximate clock speeds of that ES, but little more.

Yep, the i9-9900K is something like 15% ahead of the R7 2700X in Cinebench, so either way you slice it AMD have improved like-for-like performance by at least that much. There is definitely a question of how much of that improvement only applies to heavily multithreaded tasks - we already know Zen scales better than Coffee Lake (up to 16 threads at least) so we don't know if each core is really 15% faster, yet.

 

[KentMan]

I know, I just feel like pointing out his nonsense sometimes.


Yep, the i9-9900K is something like 15% ahead of the R7 2700X in Cinebench, so either way you slice it AMD have improved like-for-like performance by at least that much. There is definitely a question of how much of that improvement only applies to heavily multithreaded tasks - we already know Zen scales better than Coffee Lake (up to 16 threads at least) so we don't know if each core is really 15% faster, yet.

Didnt everyone speculate after the Epyc stuff last year that it was roughly 15% or so ahead of the previous Zen stuff? This would correlate with what we have just seen as well. So if it was say a 15% improvement in performance, that ES could have been running at a similar speed 4.3ghz or so... and if theres room left in the tank for higher clocks... well....

What we all really need to know is what speed that CPU was running at lol, its the 1 billion dollar question now.

 

[KillBoY_UK]

Well I can't wait to June+ to replace my main system so its now a case do I get a 2600 or a 2700 and then upgrade to a 3700 or something

 

[KungFuSpaghetti]

you can have a superior product all you want. if someone has a similar strong product at £500 you dont sell a better product at £200. anyone who believes that is in cuckoo land. thats why all cpus gpus are priced accordingly.

Agreed, which makes me think that it will be priced in that bracket but I cannot honestly believe that this will perform as well in a gaming scenario as the 9900k, although I believe the 12 or 16 core higher binned units might.

 

[Murphy]

Now if we put 2 of those chips on a package the size of ryzen thats a lot of heat we need to dissipate, they could be clocked lower to keep thermals in check, we just dont know.

You don't really need to worry about heat at a package level, within reason, it's the localised heat (the heat generated within each CCX/Core) that causes issues.

 

[subbytna]

There is a reason they didn't do that. Same as FPS counter in division......

Single core test would have measured IPC

Ubisoft fella clearly stated that now developers are looking at multicore for games. The single core IPC thing is soon to be a distant memory. When they are now going to utilise 8, 12, 16 or more cores in games, that's progress.

 

[subbytna]

What is that reason oh wise one?

Careful now, he's fishing.

 

[easyrider]

Ubisoft fella clearly stated that now developers are looking at multicore for games. The single core IPC thing is soon to be a distant memory. When they are now going to utilise 8, 12, 16 or more cores in games, that's progress.

Depending on whether the res you play is GPU bound..

At 4K 2080ti 4.2ghz Ryzen no slower than 5ghz 9900k

1080p is a different story.

 

[DragonQ]

Agreed, which makes me think that it will be priced in that bracket but I cannot honestly believe that this will perform as well in a gaming scenario as the 9900k, although I believe the 12 or 16 core higher binned units might.

It looks very likely that 12 and 16 core variants will exist, which prevents the R7 2700X's successor being much more highly priced, otherwise Threadripper chips would end up being cheaper than their AM4 counterparts. You can't have a $500 8c/16t chip because you'd then need your 12c/24t chips to start at $599 or more, and the current Threadripper 2920X is already only $649. The R7 2700X replacement might go a bit above the current $329, say to $379, but there's no way it's going to be $499 like the i9-9900K.

 

[Alexrose1uk]

Hopefully we'll see the range out by May, in time for the 50th anniversary.

 

[subbytna]

Depending on whether the res you play is GPU bound..

At 4K 2080ti 4.2ghz Ryzen no slower than 5ghz 9900k

1080p is a different story.

Not many people buying a 1300 quid 2080Ti to play at 1080p so for gaming the ryzen 2700x, at 4k, looks the better buy over intel for 4K resolution then?

 

[easyrider]

Not many people buying a 1300 quid 2080Ti to play at 1080p so for gaming the ryzen 2700x, at 4k, looks the better buy over intel for 4K resolution then?

Well people with £1200 to spend on a 2880 ti usually want the fastest chip regardless. Saving £200 getting a 2700x over a 9900k is not really on peoples minds with those budgets on new builds.

Thats one of the reasons I went for the 9900k over the 2700x ...It was only £200 more but and the chance of 5ghz on all cores...Its swings and roundabouts.

There is no perfect cpu as everyone has different needs.

 

[subbytna]

But just think you've got 6 months to feverishly post about your 9900k