6 Core 1600X 50% Faster than 7600k in CPUz Benchmark

[AliG]

The AMD Ryzen leaks and madness doesn't stop. Today some CPU-Z benchmark results have surfaced of what seems to be a six-core Ryzen 5 1600X processor. The 6 core and 12 threaded processor would have a base clock of 3.3 GHz with a 3.7 GHz Turbo. And the performance seems pretty nice.

The single threaded (single core) performance of this Ryzen unit shows 1888 points. That is precisely on par with Core i7 6850K single core performance and that I did not expect (thus that is very positive). We know that the Intel Core i7-6850K clocks in at 6/12 (cores/threads) 3.6 GHz / 3.8 GHz (base clock / turbo clock). So compared 1:1 that is really fricken close on the per core performance. The Ryzen muli-threaded performance shows 12544 points, which is faster then a Core i7-6850K who reaches 10872 points in our own testing.

While it is not a fair comparison (4 versus 6 cores), we do expect these two processors to be priced in the same range and thus compared to a quad core Core i5 7600K the Ryzen R5 offers 50% more perf then the 8314 points that CPU scores. Obviously the 7600K has much faster turbo's hence that single threaded core performance locks in at a faster 2135 points.

http://www.guru3d.com/news-story/6-c...-i5-7600k.html

Now the source is Guru3D, so I do take it with a grain of salt. But these numbers do jive with the 1700X leaks, so I think the 1600X is starting to look like a really nice value buy for midrange customers.

[BenchZowner]

The MT score is a little bit way too high for a 6core at 3.4GHz

Buttload of salt

[AliG]

The MT score is a little bit way too high for a 6core at 3.4GHz

Buttload of salt

I believe quite a few of those were sent around during the Bulldozer rumors

[xlink]

I'd believe it.

It's a synthetic. This is especially the case if the same program was run a few times and Ryzen's ANN branch predictor kicked into effect.

I don't expect it to be 50% faster in non-synthetics unless there's more parity on the clock speed front.

[vario]

I'd believe it.

It's a synthetic. This is especially the case if the same program was run a few times and Ryzen's ANN branch predictor kicked into effect.

I don't expect it to be 50% faster in non-synthetics unless there's more parity on the clock speed front.

No way in hell.
This score is either from 8 core Ryzen or is just a fabrication
First, the screens with Ryzen CPU visible are from Chinese OS, and thats quite believable, the screen with the scores however has nothing to show in termz of "this is zen" and is from a different ENGLISH OS.
Second, my Haswell-E setup at 4.2ghz 3.4NB 2666Dram cas 15-1T has a score of 1912/12003.
So my money is on "someone has fun with people" .

[AliG]

If it is truly 8-core Ryzen, isn't that score still really impressive to keep up with a 4.2 GHz Haswell?

[cdawall]

No way in hell.
This score is either from 8 core Ryzen or is just a fabrication
First, the screens with Ryzen CPU visible are from Chinese OS, and thats quite believable, the screen with the scores however has nothing to show in termz of "this is zen" and is from a different ENGLISH OS.
Second, my Haswell-E setup at 4.2ghz 3.4NB 2666Dram cas 15-1T has a score of 1912/12003.
So my money is on "someone has fun with people" .

My broadwell-e beats those scores at 4.5, I imagine this chip which offers XFR is in the XFR range and pushing a lot closer to the 4.5-5ghz range than it is 3.4.

[vario]

My broadwell-e beats those scores at 4.5, I imagine this chip which offers XFR is in the XFR range and pushing a lot closer to the 4.5-5ghz range than it is 3.4.

And you really believe that XFR overclocks the chip more than a gigahertz on all cores, yet somehow doesnt do that for single thread (it would have to achieve higher ST score if it was 6 cores instead of 8)
And also dont find it odd that screenshots showing device manager and cpuz are from chinese OS and the score shot is from english OS and NOT showing any cpuz/device manager info ?
I call BS still .But We will see.

[sergiu]

If real, then still interesting: 1888@3.7 and 12544@3.3 would mean ~1.25 scaling factor for HT. I remember I saw 1.2-1.25 in programs for P4 Northwood only after years of software optimizations, not at beginning.

[cdawall]

And you really believe that XFR overclocks the chip more than a gigahertz on all cores, yet somehow doesnt do that for single thread (it would have to achieve higher ST score if it was 6 cores instead of 8)
And also dont find it odd that screenshots showing device manager and cpuz are from chinese OS and the score shot is from english OS and NOT showing any cpuz/device manager info ?
I call BS still .But We will see.

I was saying the chip was boosted the entire time...single threaded being clocked around 4.5 and multithreaded could be similar if not a couple 100mhz lower. From how I am reading the workings of XFR it is no different than setting a boost clock for the 480. Crank the power up to +50% and set the voltage and let it scream.

As for the OS I noticed the same and have been trying to recreate that score and it looks like an 8 core haswell-e at 3.6 gets about that.

[xlink]

No way in hell.
This score is either from 8 core Ryzen or is just a fabrication
First, the screens with Ryzen CPU visible are from Chinese OS, and thats quite believable, the screen with the scores however has nothing to show in termz of "this is zen" and is from a different ENGLISH OS.
Second, my Haswell-E setup at 4.2ghz 3.4NB 2666Dram cas 15-1T has a score of 1912/12003.
So my money is on "someone has fun with people" .

synthetics can show one uarch being an order of magnitude faster(let's say factor of 3) than another for no real reason... which is why they shouldn't be taken too seriously for measurements of real world performance.