AMD Threadripper 1950X Specs leaked

[Sir Asvald]

Now, bear in mind that this is from http://wccftech.com

Quote

In terms of performance, the Ryzen Threadripper 1950X scored 4167 points in single core test. The specifics are 4081 Integer, 4025 Floating point and 4236 memory benchmark scores. Moving on to the multi-core performance, the chip scored a total of 24539 points. The specifics in this field were 28409 Integer, 33993 Floating point and 5007 memory benchmark scores. This shows that we are looking at some very high-end performance numbers from the Ryzen Threadripper series.

 

 

 

 

Quote

For comparison, we added an Intel Xeon E5-2697A V4 processor to the mix. The chip features total of 16 cores, 32 threads and a base clock of 2.6 GHz which boosts up to 3.6 GHz. It can be seen that the chip isn’t a direct comparison as it has lower base clock frequency but boost clock does match the Threadripper speeds. A single Xeon 16 core chip scores 3651 points in single core performance and 30450 points in Multi-core performance. You can note that while single core scores lacks when compared to 1950X, the multi-core performance is a much higher.

 

 

AMD Threadripper 1950X 16 Core vs Intel Xeon E5-2697A V4 16 Core Multi-Core Performance:

 

 

 

MD Ryzen Threadripper 1950X CPU Performance:

 

 

 

Looks very promising. 

 

EDIT: 

 

http://wccftech.com/amd-ryzen-threadripper-1950x-specs-performance-leak/

[Bouzoo]

5 minutes ago, Abdul201588 said:

Looks very promising. 

Hold your horses. Not that it looks good or not, geekbench is the source, which we've seen can have numbers all around the place, and then some. Need. More. Sources. 

[leadeater]

Do people use geekbench on purpose just to fuck with us? Instead of using an actually useful benchmark and leaking that we get this meaningless bunch of numbers argh!!

[Sir Asvald]

3 minutes ago, leadeater said:

Do people use geekbench on purpose just to fuck with us? Instead of using an actually useful benchmark and leaking that we get this meaningless bunch of numbers argh!!

 sorry. 

[Notional]

The weird disparity between the single and multicore score as well as the difference in clock speeds, makes these numbers utter useless. They don't make any sense.

[Sir Asvald]

Like I said guys. Its http://wccftech.com   

[leadeater]

Just now, Abdul201588 said:

 sorry. 

Hey it's not your fault, it's the source I blame.

[samcool55]

Well the ram only runs at 2133Mhz so i believe it has some potential left.

[leadeater]

http://browser.primatelabs.com/v4/cpu/search?dir=asc&q=ryzen&sort=created_at

[Notional]

1 minute ago, samcool55 said:

Well the ram only runs at 2133Mhz so i believe it has some potential left.

Good eye. Yeah, the ram is faster on the Intel platform (2400mhz). We know Ryzen benefits a LOT with faster ram, so it should perform a lot better with faster ram.

[Sir Asvald]

Just search "threadripper" and this is what I got:

 

Single core scores. 

 

[yian88]

something is not right with that test, threadripper rips intel a new one in single core but losses substantially in every multicore test, might need more optimizations there or threadrippers interfabric using 2 different dies from AMD actually suck, judging by the scores the threadripper should score low 35k+ on that bench yet it does 25k multicore wtf?

[samcool55]

1 minute ago, Notional said:

Good eye. Yeah, the ram is faster on the Intel platform (2400mhz). We know Ryzen benefits a LOT with faster ram, so it should perform a lot better with faster ram.

Yeah throw in some 3200Mhz ram and the score might go up by 10-15%, closing the gap quite a lot

 

Anyway, in the end it's the price that will decide who is the winner.

[leadeater]

Just now, yian88 said:

something is not right with that test, threadripper rips intel a new one in single core but losses substantially in every multicore test, might need more optimizations there or threadrippers interfabric using 2 different dies from AMD actually suck, judging by the scores the threadripper should score low 35k+ on that bench yet it does 25k multicore wtf?

It's multicore score is the same as Ryzen 7 so we can confidently say that these results shown here are so out of wack it's not funny.

[Notional]

Just now, samcool55 said:

Yeah throw in some 3200Mhz ram and the score might go up by 10-15%, closing the gap quite a lot

 

Anyway, in the end it's the price that will decide who is the winner.

Could be more on TR, as you have 2 dies that are connected with infinity fabric as well. That cross die communication might be even more dependent on ram speed. But it's impossible to say yet. The reviews should be interesting.

[Sir Asvald]

1 minute ago, leadeater said:

It's multicore score is the same as Ryzen 7 so we can confidently say that these results shown here are so out of wack it's not funny.

I've posted an image above. Take a look, it's a number of different tests. 

[leadeater]

Just now, Abdul201588 said:

I've posted an image above. Take a look, it's a number of different tests. 

Yea if you search ryzen the first page sorted by highest multicore results are all over 30k, you have to go in 36+ pages to get to Ryzen 7 multicore scores the same as these TR ones.

 

I have no idea how reliable those results are especially sorting by highest but TR which is essentially two Ryzen 7's should actually be able to score a statistically significant amount higher.

[SCHISCHKA]

i wonder how much money one makes by making fake benchmarks on unreleased hardware?

hypothetically.

[mariushm]

yeah, I always find it funny how they compare a system with 16 GB of memory to one with 128 GB of memory.

It's either fake, or dual channel (2x8GB) or intentionally made to look bad with 4 x 4 GB ddr4 sticks

 

My bet is fake.. it's not hard to edit a html page and then take a screenshot.

[PCGuy_5960]

Geekbench is shit. It's literally the worst benchmark you can use!

[PCGuy_5960]

9 minutes ago, Tedny said:

People, it can be fake, because it dual-slot memory benchmarks 

6 minutes ago, mariushm said:

My bet is fake.. it's not hard to edit a html page and then take a screenshot.

Or you can just use paint

[TidaLWaveZ]

From 0 - Confirmed Intentionally fake benchmark produced to defame AMD in 6 seconds.

 

I wish the masses would question all main stream news like we question our leaked info, Then the MSM wouldn't get away with shit.

[The Benjamins]

10 minutes ago, PCGuy_5960 said:

Geekbench is shit. It's literally the worst benchmark you can use!

Agree if you take the 4000 score for single thread and times it by 8 cores you get 32,000, that is 8000 less then what it got and it is a 16 core part. also it shows a higher single core part being slower in multi then a same core count lower single core scoring part. lol

 

It has the same multi score as a 1800X LOL, the single score is a little better on the 1800X

https://browser.primatelabs.com/v4/cpu/search?q=threadripper

https://browser.primatelabs.com/v4/cpu/search?utf8=✓&q=1800x

[Damascus]

5 minutes ago, TidaLWaveZ said:

defame AMD

Not AMD, just the asshats who would screw around with the benchmarks

[Bananasplit_00]

24 minutes ago, leadeater said:

It's multicore score is the same as Ryzen 7 so we can confidently say that these results shown here are so out of wack it's not funny.

id say they are fake, they looked just like the 1800X results so its probably just that with some name changeing trickery