[updated] Ryzen benchmark leaked

[Lord Nicoll]

1 hour ago, SCHISCHKA said:

Look up openbenchmark.org Linux kernel compile. It's kinda hard to navigate that site but they keep a list of comparative times across CPU. Last time I looked they only had the 8370 and 8350, I did not see the 9590 on there. From memory the fastest CPU was a Xeon with i7 and 8370 taking similar times but the interesting part is how much cheaper the fx is compared to the i7 and xeonsXeon

Well, Xeons cost more because of the dual, quad and octo socket capabilities, but that is exactly why I picked an FX-6300 for my friend who was on a budget and works with Linux and VMs a lot, for €120, he wouldn't have even gotten an i3

[Majestic]

These look...weird. 

[vitor_cut]

And in OC3D site we have this numbers.

I might be wrong be this numbers are better than the article that op post. specialy the single treaded baseline.

 

https://www.overclock3d.net/news/cpu_mainboard/amd_ryzen_7_1700x_cpu_benchmarks_leak/1

 

Spoiler

AMD Ryzen 7 1700X CPU benchmarks leak

 

New benchmarks have emerged from an AMD Ryzen CPU that is believed to be the 8-core R7 1700X, giving PC builders what could be our first look at the performance of AMD's new Ryzen architecture. 

 

Before looking at these numbers a few things need to be established, first that this is an engineering sample and not a final CPU, second that this system was tested using a low-end MSI A320 Ryzen motherboard with 2400MHz memory, all of which can negatively affect these benchmark results in the charts below.  

 

One other thing to note is that in this test CPU core Turbo mode as not detected, so it could potentially be inactive during this test.  

 

The timing on the DDR4 memory are particularly bad, sitting with fairly loose timings of (17-17-17-39 2T), which combined with a rather low speed of 2400MHz could have a notable effect on this CPU's performance, especially when compared to X99 Intel's CPUs that run with quad-channel memory. 

 

Compare this to the i7 6800K in the tables below, that is known to be using quad-channel 3200MHz memory with timings of 16-18-18-38. This means that this is far from an apples-to-apples comparison, though certainly results that are worth looking at. 

 

 

  

  

These tests are from CPUMARK, which tests several aspects of CPUs from single-core performance, integer math, floating-point performance and many more forms of calculations. Here we have 9 different CPU-related tests on AMD's Ryzen CPU engineering sample as well as data from a lot of recent Intel CPUs like the 7700K and the i7 6900K. 

When looking at the Integer Math score we can see that AMD's Ryzen architecture can really pack a punch, with AMD's engineering sample actually besting Intel's i7 6900K, which is a highly impressive result for AMD, especially given the fact that this CPU is not the Ryzen flagship model. 

  
  

 

Moving down to prime numbers we see AMD fall behind here, though Ryzen is still enough to beat Intel's i7 7700K, though it is worth noting that that CPU is a quad-core and not an 8-core. Here Intel has a clear advantage, 

In the compression test AMD's Ryzen CPU is able to compete well with Intel's i7 6900K and i7 6850K, which is grat news given AMD's rumoured pricing and the fact that this is not the rumoured 3.6/4GHz Ryzen flagship.  
  

  

In Single threaded performance, this Ryzen CPU sample is shown to have similar performance to Intel's i7 6950X, falling a little behind Intel's higher clocked CPUs. Even so, this single-core performance is much higher than anything that AMD has offered previously, which is a fantastic change for the company. 

  

  

 

Moving to Floating point and Compression performance and we can see that this Ryzen CPU sample offers performance that is in-between Intel's i7 6850K and i7 6900K, which is fantastic levels performance given Ryzen's leaked pricing. While it seems like the best per-core performance is still provided by Intel, it looks like AMD will be offering a lot of value with Ryzen, which is something that Intel should be concerned about.   

  

 

Looking at these results we can see that both AM and Intel CPUs have their strong points, though it seems like this is far fro the best that Ryzen CPUs can offer. It will be interesting to see how AMD's higher clocked Ryzen processors will perform, especially when combined wit faster memory.  

 

[dalekphalm]

29 minutes ago, vitor_cut said:

And in OC3D site we have this numbers.

I might be wrong be this numbers are better than the article that op post. specialy the single treaded baseline.

 

https://www.overclock3d.net/news/cpu_mainboard/amd_ryzen_7_1700x_cpu_benchmarks_leak/1

 

  Reveal hidden contents

AMD Ryzen 7 1700X CPU benchmarks leak

 

New benchmarks have emerged from an AMD Ryzen CPU that is believed to be the 8-core R7 1700X, giving PC builders what could be our first look at the performance of AMD's new Ryzen architecture. 

 

Before looking at these numbers a few things need to be established, first that this is an engineering sample and not a final CPU, second that this system was tested using a low-end MSI A320 Ryzen motherboard with 2400MHz memory, all of which can negatively affect these benchmark results in the charts below.  

 

One other thing to note is that in this test CPU core Turbo mode as not detected, so it could potentially be inactive during this test.  

 

The timing on the DDR4 memory are particularly bad, sitting with fairly loose timings of (17-17-17-39 2T), which combined with a rather low speed of 2400MHz could have a notable effect on this CPU's performance, especially when compared to X99 Intel's CPUs that run with quad-channel memory. 

 

Compare this to the i7 6800K in the tables below, that is known to be using quad-channel 3200MHz memory with timings of 16-18-18-38. This means that this is far from an apples-to-apples comparison, though certainly results that are worth looking at. 

 

 

  

  

These tests are from CPUMARK, which tests several aspects of CPUs from single-core performance, integer math, floating-point performance and many more forms of calculations. Here we have 9 different CPU-related tests on AMD's Ryzen CPU engineering sample as well as data from a lot of recent Intel CPUs like the 7700K and the i7 6900K. 

When looking at the Integer Math score we can see that AMD's Ryzen architecture can really pack a punch, with AMD's engineering sample actually besting Intel's i7 6900K, which is a highly impressive result for AMD, especially given the fact that this CPU is not the Ryzen flagship model. 

  
  

 

Moving down to prime numbers we see AMD fall behind here, though Ryzen is still enough to beat Intel's i7 7700K, though it is worth noting that that CPU is a quad-core and not an 8-core. Here Intel has a clear advantage, 

In the compression test AMD's Ryzen CPU is able to compete well with Intel's i7 6900K and i7 6850K, which is grat news given AMD's rumoured pricing and the fact that this is not the rumoured 3.6/4GHz Ryzen flagship.  
  

  

In Single threaded performance, this Ryzen CPU sample is shown to have similar performance to Intel's i7 6950X, falling a little behind Intel's higher clocked CPUs. Even so, this single-core performance is much higher than anything that AMD has offered previously, which is a fantastic change for the company. 

  

  

 

Moving to Floating point and Compression performance and we can see that this Ryzen CPU sample offers performance that is in-between Intel's i7 6850K and i7 6900K, which is fantastic levels performance given Ryzen's leaked pricing. While it seems like the best per-core performance is still provided by Intel, it looks like AMD will be offering a lot of value with Ryzen, which is something that Intel should be concerned about.   

  

 

Looking at these results we can see that both AM and Intel CPUs have their strong points, though it seems like this is far fro the best that Ryzen CPUs can offer. It will be interesting to see how AMD's higher clocked Ryzen processors will perform, especially when combined wit faster memory.  

 

That is a much more objective analysis of the "leaked" figures.

 

Everyone is looking at that, then ignoring the fact that a 3.4GHz CPU is going up against CPU's with as high as a 4.7GHz OC.

 

Whoever ran these benches either didn't have all the CPU's on hand (and thus picked out already submitted results), or intentionally made Ryzen look as bad as possible.

 

I assume it's the former, and that they simply needed to use pre-existing results to compare.

[TorqueS]

Single core performance score of a 7700K at 4.2GHz is 27 points higher than a 5820K at 3.3GHz and a 5960X at 3GHz beats by almost 100points a 6900K at 3.2GHz???!!!
Based on what we know of IPC improvements in Intel's chip lineups, this...either is a load of bs...or a very bad benchmark.
Since we're getting closer to the release of Ryzen, a lot of benchmarks like this will appear. I will wait for trusted reviewers to show benches.

[Humbug]

39 minutes ago, dalekphalm said:

Everyone is looking at that, then ignoring the fact that a 3.4GHz CPU is going up against CPU's with as high as a 4.7GHz OC.

ya and also people are drawing conclusions forgetting that this CPU had boost disabled. So it's stuck at base clock unlike all the others...

[dalekphalm]

13 minutes ago, TorqueS said:

Single core performance score of a 7700K at 4.2GHz is 27 points higher than a 5820K at 3.3GHz and a 5960X at 3GHz beats by almost 100points a 6900K at 3.2GHz???!!!
Based on what we know of IPC improvements in Intel's chip lineups, this...either is a load of bs...or a very bad benchmark.
Since we're getting closer to the release of Ryzen, a lot of benchmarks like this will appear. I will wait for trusted reviewers to show benches.

So, if you read the details closely, you'd see a few things.

 

First:

The 7700K is OC'd to 5GHz

The 5820K is OC'd to 4.8GHz

The 5960X is OC'd to 4.7GHz, and finally,

The 6900K is OC'd to 4.2GHz

 

The graph was TERRIBLY made, but all of that info is sthere at the beginning of the article.

 

The Ryzen chip is also locked at 3.4GHz, so even the slowest OC'd chip (6900K) is still running 800MHz faster, with the 7700K running a full 1.6GHz faster.

 

This is a very very bad comparison, since people won't look at the details, and will come to the wrong conclusions.

 

[Doobeedoo]

All over the place these numbers... Whatever though. We'll see the real results well soon enough. Top tier 4, 6, 8 core SKU benchmarks, OCs from retail chips is all we need.

[Revinval]

Its beats the old AMD CPU and its clocked 1.7GHz lower hahaha and considering the other chips are also overclocked all to hell and it stays on their heels this while terrible on the surface is actually pretty decent news. Just gotta wait a bit for real benchmarks but the real danger for Intel's mass market will be the 4/6 core variants. Since a 6c/12t is more of consumer I7 killer (if it can keep up per core)

[Dietrichw]

3 hours ago, dalekphalm said:

So, if you read the details closely, you'd see a few things.

 

First:

The 7700K is OC'd to 5GHz

The 5820K is OC'd to 4.8GHz

The 5960X is OC'd to 4.7GHz, and finally,

The 6900K is OC'd to 4.2GHz

 

The graph was TERRIBLY made, but all of that info is sthere at the beginning of the article.

 

The Ryzen chip is also locked at 3.4GHz, so even the slowest OC'd chip (6900K) is still running 800MHz faster, with the 7700K running a full 1.6GHz faster.

 

This is a very very bad comparison, since people won't look at the details, and will come to the wrong conclusions.

 

WCCF actually did a good thing and remade the graphs using data entries from chips at stock speeds. The 5960X and the 8350 were the only ones not on windows 10 .  http://wccftech.com/amd-ryzen-7-1700x-389-8-core-cpu-benchmarks-leaked/

 

http://cdn.wccftech.com/wp-content/uploads/2017/02/AMD-Ryzen-7-1700X-CPU-PassMark.jpg  Couldn't get image to work with the URL

[dfsdfgfkjsefoiqzemnd]

12 minutes ago, Dietrichw said:

The 5960X and the 8350 were the only ones not on windows 10

That makes those results invalid right away.  OS overhead can have a rather substantial influence on benchmarks. 

[Dietrichw]

1 minute ago, Captain Chaos said:

That makes those results invalid right away.  OS overhead can have a rather substantial influence on benchmarks. 

Still doesn't ruin the whole table, even if you throw those out (not like the 8350 mattered) the tables are more accurate than the OC results

[leadeater]

8 hours ago, Dietrichw said:

WCCF actually did a good thing and remade the graphs using data entries from chips at stock speeds. The 5960X and the 8350 were the only ones not on windows 10 .  http://wccftech.com/amd-ryzen-7-1700x-389-8-core-cpu-benchmarks-leaked/

 

http://cdn.wccftech.com/wp-content/uploads/2017/02/AMD-Ryzen-7-1700X-CPU-PassMark.jpg  Couldn't get image to work with the URL

Damn, those SSE scores.

[PlayStation 2]

Isn't A320 a low end chipset for Ryzen?

Something seems really off. Why would they test with a low end chipset?

[Dietrichw]

2 minutes ago, Dan Castellaneta said:

Isn't A320 a low end chipset for Ryzen?

Something seems really off. Why would they test with a low end chipset?

That's what the person who leaked it had access to. It is hard to get pre release hardware without NDA's

[Dabombinable]

3 minutes ago, Dan Castellaneta said:

Isn't A320 a low end chipset for Ryzen?

Something seems really off. Why would they test with a low end chipset?

Chipsets are chipsets now.....its not like the chipsets for Ryzen will be designed and manufactured by VIA or SIS.

[dalekphalm]

4 minutes ago, Dan Castellaneta said:

Isn't A320 a low end chipset for Ryzen?

Something seems really off. Why would they test with a low end chipset?

Yes A320 is a lower end chipset, w/ no OC ability. Kind of an odd chipset to pair with, but it's possible whoever did the benchmarks could only get their hands on that one board.

[PlayStation 2]

1 minute ago, Dabombinable said:

Chipsets are chipsets now.....its not like the chipsets for Ryzen will be designed and manufactured by VIA or SIS.

I know, but I'm kind of confused as to why they'd use a low end board.

[PlayStation 2]

2 minutes ago, dalekphalm said:

Yes A320 is a lower end chipset, w/ no OC ability. Kind of an odd chipset to pair with, but it's possible whoever did the benchmarks could only get their hands on that one board.

 

4 minutes ago, Dietrichw said:

That's what the person who leaked it had access to. It is hard to get pre release hardware without NDA's

Fair enough. Just seemed kinda confusing as to why that'd be the only thing they'd have access to.

[Dabombinable]

2 minutes ago, Dan Castellaneta said:

I know, but I'm kind of confused as to why they'd use a low end board.

It could be to demonstrate that you wouldn't need a high end motherboard.


BTW-the benchmark is available for direct comparison. I'm going to compare to my 4790K at 4.3GHz, my Phenom II P920 (1.6GHz flat) and A8 4555M (1.6GHz base, 1.8GHz both modules, 2.4GHz single threaded).

[i_build_nanosuits]

OHH fuck...CPU single-threaded performance slower than stock 5820K 3.3ghz...

 

[dalekphalm]

3 minutes ago, i_build_nanosuits said:

HOW fuck...CPU single-threaded performance slower than stock 5820K 3.3ghz...

 

You... didn't actually read the article did you? There were like... 3 or 4 posts at least, that explained this. The 5820K was OC'd to 4.8GHz. The graph for some reason lists the stock base (not even turbo) speed, but it's clearly stated that each chip was clocked to a certain point (and lists each OC).

[Dietrichw]

3 minutes ago, i_build_nanosuits said:

HOW fuck...CPU single-threaded performance slower than stock 5820K 3.3ghz...

 

 

If you look at the image above the tables it shows the chips are overclocked from what is shown on the tables. That 5820K is at 4.00-4.80GHz. WCCF remade the tables from runs of stock chips. IDK why they chose to compare it to OCed data.

[i_build_nanosuits]

1 minute ago, dalekphalm said:

You... didn't actually read the article did you? There were like... 3 or 4 posts at least, that explained this. The 5820K was OC'd to 4.8GHz. The graph for some reason lists the stock base (not even turbo) speed, but it's clearly stated that each chip was clocked to a certain point (and lists each OC).

and the clocks of the ryzen were?!

cause there is no way that thing beat a 4.8ghz 5820k unless it,s heavily overclocked too...

[dalekphalm]

1 minute ago, i_build_nanosuits said:

yeah well the ryzen was clocked to 4.7ghz then if the 5820k was 4.8...right? that what it says which is not really any better?

Err no. Again. If you had read the article, you'd clearly see that it states the Ryzen chip was clocked at 3.4GHz with Turbo disabled. Besides, the A320 chipset isn't even capable of OC'ing the CPU to 4.7GHz, had they even wanted to.