Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Feb 17, 2017.
Those leaks are being released in such an orderly fashion. :3eyes:
Interesting that the 6 core seems to have the full 16MB L3. I wonder if that makes it have a tad better IPC than the 8 core in cache-sensitive operations.
WOW! FX 6300 really kick ass for that 90€ when I bought it, 2500K was ~230-250€ as I remember.
I made the math in a minute because i had to go so i didn´t remember that the speed difference is that big.
But then it means that Ryzen has the same IPC has Kaby Lake, like buhehe already said, and that seems to good to be true, at least to me...
Things are getting interesting, that´s for sure.
Oh boy! Now if they price them around $200-$250 max they will sell like hotcakes in my opinion. Looks hella-good cant wait to see how they do in gaming benchmarks.
3.20GHz Base, 3.50GHz Turbo for all Threads, 4.00GHz 1 to 4 Threads (1 per module)
If this all goes to plan for AMD (and us), even the 1400X 4c/8t could be a great budget gaming CPU right now especially @ 4.5GHz +
Yeah, my old FX6300 had the same L3 cache as the FX8xxx
Now that I have a vague idea of the prices, I think I will settle on a 6700k + Z170 pro gaming + 16GB combo which I am getting for a killer price, and much lower than if I buy a Zen CPU with new mobo and RAM.
So hopefully I don't regret it... :nerd:
I suspect the CPU-Z score is a FAKE.
My 6900K with a fixed clock of @ 3.9Ghz and 6 of the 8 cores enabled in BIOS obtains a score of 1878/11427 points, resulting a multi-thread gain of x6.084. The Ryzen X1600 score is 1888/12544 points, with a multi-thread gain of x6.644.That is just wrong. If the CPU was using Turbo mode, then the single thread score should be much higher, and the multi-thread gain lower than x6, since the CPU is using a higher single-core frequency than with multi-core.
Even with an identical single and multi-thread frequency the multi-thread score is way disproportioned, much higher than the single-thread score should dictate… Unless this is an 8 core CPU and they just run the test with 12 threads enabled. This way the benchmark runs automatically on all 8 cores, leaving just a few unused hyper-threaded virtual cores. That would explain the strange results.
My 4770K @ 4,5 GHz.
Wouldn't that depend on how AMD's turbo implementation works?
For reference , 3570K @ 4.6Ghz
Those 2 screenshots were taken on different computers. Look at the window buttons on the top-right corner of each window, the shell theme is different. I believe the top photo is Windows 10 and the bottom one is Windows 8. That doesn't automatically mean the scores are fake, but it certainly does nothing to prove the scores are from the Ryzen CPU depicted on the second screenshot. I guess they could have upgraded the OS between taking the 2 screenshots, or have 2 different Ryzen machines on hand running a different OS.
You are wrong. It all depends on how they implemented the Hyper Threading feature.
As an example, my old dual Xeon x5670 gets
1153 single thread
14632 multi thread (12 cores and 24 threads)
for a ratio of 12.69
2297 single threaded and 10323 multi threaded for me, Ryzen is gonna be nice! Won't upgrade to it though, will wait for the next tick/tock.
we don´t know all details about Turbo, maybe that model is @3.30GHz base 3.50GHz for all threads and 3.70GHz for half threads or 3.30GHz base 3.70GHz turbo for all threads and a bit more for 1 thread (XFR, eXtended Frequency Range)?
CPUz bench is bad on multithreaded non optimized OS, check CPUz bench results before and after optimization with my FX-8320E
The ratio is better if you consider Turbo Boost clocks for mono and multithreaded, if your CPU is @2.66GHz base, Turbo for all cores could be 2.80GHz, Turbo for half cores 2.93GHz and 3.07GHz for 3 or less cores, example:
or your own model
http://www.cpu-world.com/CPUs/Xeon/Intel-Xeon X5670 - AT80614005130AA (BX80614X5670).html
then you can divide mono and multi core score by respective clock
1153/3.33GHz = 343.93
14632/3.20GHz = 4752.5
4752.5/343.93 = 13.818x