AMD Ryzen 9 3950X beats Intel i9-10980XE in 3DMark Firestrike by almost 25%
As releases dates are getting closer, we'll see more and more infor leak somehow. New is a 3DMark Firestrike for the Ryzen 9 3950X, which can be compared with Cascade Lake-X. Guess who won?
We assume the obvious being that we assume these benchmarks are genuine – as provided by the ever-present CPU leaker TUM_APISAK.
The AMD Ryzen 9 3950X (16-core) result which we see here was run on a X570 motherboard and shows a Firestrike test, shows a score of 32082. The maximum turbo clock shows just over 3,700MHz and the stock clock of 3,500MHz. The AMD rig was outfitted with 16GB of Kingston DDR4 at 3,200MHz and 16 cores with 32 threads active. An Core i9-10980XE (18-core) result with 32GB of 2,666MHz Samsung DDR4 scored 25,838, is great but substantially less.
Firestrike is heavily threaded limited though, ergo compared is the CPU tied Physics score.
AMD Ryzen Threadripper 3960X, 3970X and 3990X Launch Dates Leak - 10/18/2019 07:04 PM
news reaches the web today that AMD will release three Threadripper 3000 SKUs: 3960X, 3970X and 3990X. The first two would be unveiled on November 5th, the embargo on sales and reviews would be lifted...
AMD Ryzen 7 3750X Spotted - Has a TDP of 105 watts - 10/17/2019 09:03 AM
Recently we have seen some OEM specific Ryzen 3000 announcements, there might be one more inbound as we do not see this one end up in retail, a Ryzen 7 3750X showed up in a recent product master list...
AMD Ryzen 5 3500X Benchmarks - 10/11/2019 09:53 AM
Silently AMD expanded their Ryzen 3000 line-up this week, exclusively in Chinafor the time being. Ryzen 5 3500X has been benchmarked. The non SMT Ryzen 5 is compared with Intel's Core i5 9400F - who...
Team Group Outs Memory and PCI-E Gen4 x4 M.2 SSD for Supporting AMD Ryzen 3000 and X570 - 10/08/2019 11:52 AM
Team Group releases T-FORCE DARK Z α DDR4 Gaming Memory and CARDEA ZERO Z440 PCI-E Gen4 x4 M.2 Solid State Drive, which are specially made for AMD RYZEN 3000 processor series and the latest ...
Over 100 Improvements to be included in upcoming AMD Ryzen 3000: AGESA 1.0.0.4 - 10/04/2019 10:36 AM
While everybody is moving towards AGESA 1.0.0.3ABBA to squeeze out every last drop of performance from Ryzen 3000 processors, AMD is already at work on AGESA 1.0.0.4 as unreached boost frequencies is...
Senior Member
Posts: 1585
Joined: 2006-02-18
Intel was mocking with customers and enjoying the throne for far too long ... the prices were accordingly ... now things have changed and they are SO LOST but they still have the market share that will fall in the next years if they dont come with something truly magic !
How many sockets did the make ? how many times they made customers buy new motherboards ? I think there are maybe 10-15 sockets for basically the same CPUs ... they were so greedy even the chipsests had to be sold in imense amounts .... The performance increase was at least mediocre between the many "generations" they had and it was so easy and comfy to count the money coming in with absolutely nothing new.
Take an I7 970 from 15 years ago compare it to an i7 from 5 years ago ... the difference ... marginal. All those core i5 with no hyperthreading all those K and Extreme processors that costed like a small town car ... it will all go to dust and I will be dancing in that dust in 5 years with my shiny new 64 core processor made by AMD.
Senior Member
Posts: 841
Joined: 2015-05-19
How many sockets did the make ? how many times they made customers buy new motherboards ? I think there are maybe 10-15 sockets for basically the same CPUs ... they were so greedy even the chipsests had to be sold in imense amounts ....
That argument is really getting old. I get it, there is enthusiasts here that might upgrade a CPU and keep using an old mainboard. But in the real world, the majority of Intels sales do not go to such people. Intel doesnt really care if they sell Chipset 1 or Chipset 2, because business and OEM will largely never upgrade CPUs only - and thats where the real money is for Intel. Even the majority of self-build systems will not upgrade frequent enough to ever make use of this.
Making new chipsets was largely just "easier", they had all the freedom to design the CPUs how they wanted to, and they just didn't care, because the market segment that might care about reusable boards is an extremely tiny fraction.
Its even funnier that you post in here, because the 10000 HEDT series is actually the 3rd line of CPUs that runs on the X299 platform.
Senior Member
Posts: 218
Joined: 2007-06-11
Intel was mocking with customers and enjoying the throne for far too long ... the prices were accordingly ... now things have changed and they are SO LOST but they still have the market share that will fall in the next years if they dont come with something truly magic !
How many sockets did the make ? how many times they made customers buy new motherboards ? I think there are maybe 10-15 sockets for basically the same CPUs ... they were so greedy even the chipsests had to be sold in imense amounts .... The performance increase was at least mediocre between the many "generations" they had and it was so easy and comfy to count the money coming in with absolutely nothing new.
Take an I7 970 from 15 years ago compare it to an i7 from 5 years ago ... the difference ... marginal. All those core i5 with no hyperthreading all those K and Extreme processors that costed like a small town car ... it will all go to dust and I will be dancing in that dust in 5 years with my shiny new 64 core processor made by AMD.
What the hell are you talking about "marginal" Lol are you joking, get your fanboi cancer outta here mate. What you wrote is complete BS.
Senior Member
Posts: 387
Joined: 2016-07-09
And yet the title says it's 25% faster.... This is a false and misleading piece.
Senior Member
Posts: 841
Joined: 2015-05-19
Just look up 9980XE numbers if you want something to compare to. The new CPU should if anything be slightly better (slightly better clock, hardware security mitigations), but probably not much. Obviously its impossible to judge an entire system from the benchmark page, since it has no details on cooling used, but you can extrapolate from the clock, I guess. A random 9980XE result with a 4.6GHz boost clock had around 32k Physics as well.
The Physics numbers shown in that "leak" are far too low for any real-world setup that a 10980XE will ever run it, because who buys such a thing and then doesn't give it proper cooling to stretch its legs?
TL;DR:
Benchmarks without controlled circumstances are meaningless.