AMD RyZen CPU Architecture for 2017

Au contraire, I'm betting that since Intel didn't improve basically anything with Kaby Lake, they'll struggle to reach this skl -> kbl performance increment. We are talking abount the improvement of an (already) improved process. And we don't have to speculate on the speed increments, we already know 8700k will reach 4.7 ghz turbo max 4.5 of 7700k

They'll be fine only because they've increased the core count . Also another area where Intel has a chance to improve with this generation is overcklocking.
 
Intel is still well ahead in per-thread performance, especially in games, so they don't need much of an improvement in this department. With 2 more cores they'll probably match the 1800X in most cases for well threaded applications. Which one comes out on top ultimately will be interesting to see, but with Intel's lead in games, I think Coffee Lake will be a much more appealing proposition overall. Plus, it's apparently going to be much cheaper, although I would expect big Ryzen price cuts in response.

Incidentally, this might mean ThreadRipper price cuts as well, as only the latter will be able to truly beat Coffee Lake.
 
Ryzen Mobile pictured:

1fxXMCX.jpg


0glwfqk.jpg



Seems to be around 220mm^2. The same article says Kaby Lake 4C+GT2 is 180mm^2 but it could be wrong since the 2C+GT2 is below 100mm^2.
Regardless, the iGPU is probably what's taking most of the die area. Polaris 11 (16CU) is 123mm^2 and Polaris 12 (10CU) is 100mm^2. Raven Ridge's iGPU should be around those numbers.
 
ToTTenTranz said:
The same article says Kaby Lake 4C+GT2 is 180mm^2 but it could be wrong since the 2C+GT2 is below 100mm^2.
Click to expand...
I've seen 123mm^2 for 4C Kaby Lake R (somewhere, sorry no source), so the larger number may include the "chipset" (controller hub) die.

For comparisons sake, what's the word on integrated connectivity on Ryzen Mobile?
 
Alexko said:
is about 15% ahead per thread and per clock. But does it matter whether a per-thread performance lead comes from IPC or clock speed? The end result is the same.

http://www.hardware.fr/articles/970-5/coffee-lake-kaby-lake-ryzen-3-5-ghz.html
Click to expand...
http://www.hardware.fr/articles/970-6/protocole-test.html

Looks like they are leaving Ryzen performance on the table( relative to the other benchmarked CPUs ), with memory that makes the IF a bottleneck.

15% perf per clock is not what I would have the average performance deficit more like 5% ( excluding the few things that really scale with 256 bit vectors like linpack).
 
So looks like Samsung-B die RAM prices have dropped finally to the same prices as when Ryzen was released. At the same time, with the CL CPU release, other non-b die ram is about the same price! So I managed to sell my crappy Hynix TridentZ ram on ebay for the same price I bought them at back in March, and ordered some better CL14 b-die RAM instead.
 
Is there a way to tell which memory devices your DIMM is using, like reading it out through SPD or such or can you only find out by word-of-mouth from people in the know?
 
Grall said:
Is there a way to tell which memory devices your DIMM is using, like reading it out through SPD or such or can you only find out by word-of-mouth from people in the know?
Click to expand...
One of the only programs that read what chips are present is Thaiphoon Burner. Depending on the ram, you can also determine the chips by specific codes printed on the DIMMS.
 
Got my new RAM, threw it in and no issues using D.O.C.P. It was really nice just to be able to set it to it's full speed and actually working for once.
 
Last edited:
Arnold Beckenbauer said:
Don't think so, but possible that U stands for 12-25 W and is configurable.
Click to expand...

I don't know where pcper got the 35W number from, but those early slides claimed the FP5 BGA Raven Ridge would range between 12 and 45W.

Of course for simplicity sake it would make much more sense for AMD to use the U suffix to place Raven Ridge directly against Intel's U offerings, and that would mean 15-28W.
The 35-45W versions should use a different suffix.
 
ToTTenTranz said:
I don't know where pcper got the 35W number from, but those early slides claimed the FP5 BGA Raven Ridge would range between 12 and 45W.

Of course for simplicity sake it would make much more sense for AMD to use the U suffix to place Raven Ridge directly against Intel's U offerings, and that would mean 15-28W.
The 35-45W versions should use a different suffix.
Click to expand...
https://i0.wp.com/www.techarp.com/w...n-Model-Number-Decoder-01.jpg?resize=1000,583

i think its a fair assumption that:
M is sub 15watt
U is 15-28watt
H is 28-45 watt
 
Ok, at the moment I'm really tempted to get a Dell 13 inch with the Core i5 8250u (4 cores / 8 threads) for £729. Processor has a TDP of 15W. Given that a good GPU is a nice to have, but not essential, do you think I should wait for Ryzen?
 
Isn't it a good idea to at least check out ryzen reviews first before buying anything? Unless it's an excessively long wait perhaps, or if you really need a laptop right now...
 
You must log in or register to reply here.

Similar threads

F
RDNA5 Speculation
Replies
2
Views
1K
Erinyes
E
Rys
Reopening Architecture and Products + rules and guidelines changes
3 4 5
Replies
90
Views
16K
Shifty Geezer
Shifty Geezer
A
AMD: Zen 3 Announced [2020-10-08]
10 11 12
Replies
220
Views
91K
davis.anthony
D
I
Apple A16 SoC
Replies
1
Views
3K
Entropy
E
A
  • Locked
"next generation RDNA architecture" with hardware accelerated ray tracing
2
Replies
24
Views
14K
BRiT
BRiT
Back
Top