Sharing Ryzen 1700 vs i7-6700 result

[NikosD]

Why don't you check yourself?
https://www.visualstudio.com/downloads/
I have completely abandoned any thoughts of putting again developer tools in my system.

I remember you looking for some info of different compilers than Intel's to test.
I offered you one!

Nevermind, if you are not in the mood just forget it.

[Sagittaire]

One more time, i don't think that this bench work correctly:

if you compare i3-4170 vs Rysen 7 in single core mode, i3-4170 produce better result and by far than Rysen 7 for x86, x87, SSE2, AVX or AVX2.

Anyway for single thread in x264 bench you have:

Rysen 7 1800X: 0.6175 fps/ghz
Core i7-6900K: 0.5756 fps/ghz
Core i7-5960X: 0.5857 fps/ghz
Core i7-7700K: 0.6866 fps/ghz
Core i7-4790K: 0.6363 fps/ghz
Core i7-3770K: 0.5512 fps/ghz
Core i7-2600K: 0.5052 fps/ghz

How Rysen can be really catastrophic in all synthetic test for all intructions and fight with i7-6900K in real application like x264 or x265?

http://www.hardware.fr/getgraphimg.php?id=440&n=1

[NikosD]

One more time, i don't think that this bench work correctly:

if you compare i3-4170 vs Rysen 7 in single core mode, i3-4170 produce better result and by far than Rysen 7 for x86, x87, SSE2, AVX or AVX2.

Anyway for single thread in x264 bench you have:

Rysen 7 1800X: 0.6175 fps/ghz
Core i7-6900K: 0.5756 fps/ghz
Core i7-5960X: 0.5857 fps/ghz
Core i7-7700K: 0.6866 fps/ghz
Core i7-4790K: 0.6363 fps/ghz
Core i7-3770K: 0.5512 fps/ghz
Core i7-2600K: 0.5052 fps/ghz

How Rysen can be really catastrophic in all synthetic test for all intructions and fight with i7-6900K in real application like x264 or x265?

http://www.hardware.fr/getgraphimg.php?id=440&n=1

We must not compare apples with oranges.

The synthetic test measures theoretical FPU performance using double precision FP64 numbers in FADD/FSUB/FMUL operations of legacy x87 and vector processing SIMD like SSE2/AVX/AVX2-FMA3

None of them is used by apps like x264 or x265.

So, think of x264 & x265 apps as a measurement of integer SSE2/AVX/AVX2 performance and FlopsCPU as a measurement of FP64 SIMD SSE2/AVX/AVX2 performance.

That x86 test of FlopsCPU is irrelevant.

Now, if you try to find out FP64 real world apps that behave like FlopsCPU benchmark, you have to dig in the world of HPC, scientific, 3D, rendering apps although they are not so much optimized for FP64 SIMD like that synthetic-theoretical app.

[Sagittaire]

Now, if you try to find out FP64 real world apps that behave like FlopsCPU benchmark, you have to dig in the world of HPC, scientific, 3D, rendering apps although they are not so much optimized for FP64 SIMD like that synthetic-theoretical app.

Rysen 7 seem really make good work in all 3D rendering apps too for exemple ...

There is in fact only a very rare real application where Rysen does not give a good result. Why test SIMD that are never used in practice?

[CruNcher]

Intel put the ultra efficient Mobile versions on the table and KabyLake is now Flooding the Laptop Market we gonna see soon how well Ryzen will really do as well how well AMD will GPU wise do vs the Nvidia + Intel combination ;)

At least Intel + AMDs GPU combination will follow very soon RX 5xx (560M/570M) updated version against mostly the GTX 1050 TI (14nm) updated VPU/GTX 1060 (16nm).

And intels 4 Cores/4T do really good so far :)

https://www.youtube.com/watch?v=mwiY2OOz1sY&list=UUGtSdR7VYd3U9J4O99vl_Cg

Efficiency vs the Base (PS4 Pro) seems not perfect but not that bad at all :)

https://www.youtube.com/watch?v=PVEbJzIEABg

So soon well see what AMDs 4Cores/4T and 8T brings on the table vs it at perf/watt price overall ;)

More data to crunch is incoming

https://www.youtube.com/watch?v=TxYVbg2VMuw&list=UU7D8-hkY0PrS5BNgl_HBXNg

i7-7300HQ
i7-7700HQ

are damn efficient @ 45W with integrated GPU

Zen needs to at least stay competitive in that Performance range Mobile

Bristol Ridge currently tumbles somehwere @ 65W (not efficient enough).

It seems currently AMD needs to counter attack a price range of 1000 on the lowest entry with rather better or at least equal performance to a Intel Kaby Lake (14nm) /Nvidia GTX 1050 TI (14nm) combo Platform (with both superior VPU Software/Hardware state), that fight will be a tad harder then the Desktop one every watt inefficiency vs price counts now (Software/Hardware) ;)

Biggest problem for AMD Vega is not ready neither Desktop nor Mobile and they have to go without very important improvements into this now with a updated Polaris before Raven Ridge is even ready with it's Vega Core and could gain some traction with the MultiGPU scaling.

[zub35]

Ryzen7 1700, avx2 on/off
http://rigaya34589.blog135.fc2.com/blog-entry-909.html

--

Owners of ryzen, test x265 on different compilers GCC, VS, ICC
With AVX2 enabled and disabled in the settings. To find out the fastest speed.

GCC / VS: http://x265.ru/builds
ICC: http://msystem.waw.pl/x265

[burfadel]

Ryzen7 1700, avx2 on/off
http://rigaya34589.blog135.fc2.com/blog-entry-909.html

--

Owners of ryzen, test x265 on different compilers GCC, VS, ICC
With AVX2 enabled and disabled in the settings. To find out the fastest speed.

GCC / VS: http://x265.ru/builds
ICC: http://msystem.waw.pl/x265

It doesn't show too badly for Ryzen. An important consideration is RAM speed, something that they're improving with Ryzen and hopefully they'll have resolved in the next couple of months. Ryzen apparently loves faster memory, that test was done with 2400 MHz memory on Ryzen and 3600 MHz memory on the i7-7700K. The i7 was also clocked at the maximum overclock you should consider when not using high CPU load continuously on a regular basis (which is closer to 4.6 GHz). The 1700X was at stock speeds.

I think bios and software updates would make that graph show something different in a good way for the 1700X in 6 months time, assuming they have fully updated and running the fastest realistic RAM they can (maybe like 3600 MHz).

[CruNcher]

It indeed becomes interesting comparing clock for clock with a very fast pooling time :)

https://www.youtube.com/watch?v=KDZ1dVu80as


unfortunately no one really does synced Power tests at the same time but obviously the Goal reaching 45W like KabyLake could be reached now :)

Without doubt it is one of the best AMD processors since Athlon 64 :)

And really excited to the see the Mobile results in the not so distant future vs Intel of Zen :)

[ShogoXT]

Ryzen7 1700, avx2 on/off
http://rigaya34589.blog135.fc2.com/blog-entry-909.html

--

Owners of ryzen, test x265 on different compilers GCC, VS, ICC
With AVX2 enabled and disabled in the settings. To find out the fastest speed.

GCC / VS: http://x265.ru/builds
ICC: http://msystem.waw.pl/x265

If I read that correctly, so it IS faster with avx2 off. Would avx be better than sse4.2?

Turns out my ram was only running at 2133 before. The mobo reduced it automatically instead of telling me the settings failed. Got new ones and it's running at 3200 for real this time. I was actually seeing frame lag in some games before I managed to eek it up to 2400 before changing the ram.

Ryzen loves fast ram , remember that.

[sneaker_ger]

If I read that correctly, so it IS faster with avx2 off.
For some of the tests at least. (x264 8bit and x265 8 bit but not x265 10 bit)

[CruNcher]

Seems slowly we see several Studios optimizing intensively for AMD already based on their PS4 work :)

http://media.bestofmicro.com/ext/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wLzQvNjU5NjY4L29yaWdpbmFsL2ltYWdlMDUyLnBuZw==/r_600x450.png

http://media.bestofmicro.com/ext/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wL0IvNjU5Njc1L29yaWdpbmFsL2ltYWdlMDUyLnBuZw==/r_600x450.png

http://media.bestofmicro.com/ext/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8xLzMvNjU5NzAzL29yaWdpbmFsL2ltYWdlMDUyLnBuZw==/r_600x450.png



Eidos new Glacier2 based Engine goes a complete different way then anything else currently.


Even with its GPU internally Energy Efficiency still on Intels side for system idle.

https://img.purch.com/w/711/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8zLzYvNjU4NDgyL29yaWdpbmFsLzAxLUlkbGUucG5n

But if we look at Workloads it changes rapidly

https://img.purch.com/w/711/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8zLzgvNjU4NDg0L29yaWdpbmFsLzAzLUdhbWluZy1NZXRyby1MTC1GSEQtSGVhdnktV29ya2xvYWQtLnBuZw==

https://img.purch.com/o/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8zLzMvNjU4NDc5L29yaWdpbmFsLzA0LVRvcnR1cmUtTG9vcC5wbmc=

https://img.purch.com/w/711/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS84L0MvNjU4NjY4L29yaWdpbmFsLzExLVRlbXBlcmF0dXJlcy1Ub3J0dXJlLnBuZw==


Pretty Amazing results if the data is not wrong measured and i trust Igor much more then any other Reviewer on this Entire Planet :)


That's some crazy Preview to what we might gonna see Mobile next AMD might gonna take the Lead with Raven Ridge based Mobile Systems.

Though Intel has the drawback here of the Internal GPU that shouldn't be forgotten as well ;)

And overall some of AMDs R&D still lacks in the whole package ;)

Mobile though it's already pretty clear that Intel has no chance and also the Combination of Nvidia + Intel on the lowest power possible might not save it this time it looks very good for AMD taking the Mobile Lead in overall Performance and Consumption in front of Intel + Nvidia ;)

Scorpio will be yummy :)

[NikosD]

I have updated my x264 & x265 benchmark results of this thread here:
https://forum.doom9.org/showthread.php?p=1800781#post1800781

Take a look.
It seems that RyZen has better IPC for x265 than x264 using the default settings of the benchmark, compared to older Intel architectures (Sandybridge & Haswell) but not for Skylake.

[CruNcher]

~22W more power efficient then Skylake with GPU ;)
and still ~8W then Kaby Lake also with GPU

though we have no x264/x265 based Workload Power Efficiency Data compared with AVX/AVX2.

but you could partly relate it to existing data looking at temperature results.

which are overall more efficient measurable with the OS latency inside the OS then Power ;)

we see roughly 8W higher efficiency then i7-7700 @ 3.8 GHz without any GPU core on the Gaming test

But the torture loop becomes more interesting which also displays more what x264 and x265 are ;)

and again wee see ~22W higher efficiency @ same clock then Skylake.

Though we also see that temperature wise KabyLake still leads by ~-2 °C and at 3.8 Ghz even a whoping ~-14 °C which is clear if you look @ the Power difference of 85W vs 142W almost a ~60W difference now we could calculate ~30W per real core ;)

and Ryzen over Skylake by that exact ~-2 °C

Also a thing i hate about Reviews needing to search for important data on zillion of pages instead of getting it displayed efficiently in 1 view per tested workload

All these Zillion of Pages make no real sense instead of all data per workload combined in 1 view.

Now it becomes clearer why Raven Ridge will be 4 cores only @ first even with the better yields 6 cores could become problematic overhead wise vs Intel Mobile ;)

So overall that Mobile lead could be faster over for AMD then thought with Coffe Lake ;)

[NikosD]

AMD could have a bigger problem with the new HEDT platform of Intel using X299 chipset and Skylake X/ Kabylake X CPUs than Broadwell-E HEDT.

Kabylake X is a quad core meaningless decision.

But Skylake X will go up to 10C/20T and will be much faster than the Broadwell-E HEDT CPU.

Thank god this new HEDT platform from AMD will use 16C/32T CPUs and if the rumor is true, it will eat Skylake X for breakfast.

http://www.guru3d.com/news-story/amd-x390-and-x399-chipsets-diagrams-reveal-hedt-information.html

[CruNcher]

Skylake is not Kaby Lake and Intel Plans allready to attack AMD with 6 Cores Mainstream including GPU with better Energy Efficiency then AMDs 8 Cores without GPU

The most efficient Zen combination we gonna see in Scorpio a really ambitious platform project Hardware/Software :)

Even Raven Ridge will be only a Glimpse of it's overall Efficiency.

What normal AVG PC users get sold currently is only the trash stuff ;)

[NikosD]

Nobody cares for energy efficiency on desktop or HEDT as long as the power consumption and temperatures are in sane levels.

AMD is on par with Intel on that and it has faster CPUs.

Kabylake CPU is a higher clocked Skylake CPU just to somehow compete with Ryzen 7.

That 6core CPU from Intel only exists because of Ryzen 8C/16T.

There would be no 6core CPU for mainstream desktop from Intel, if AMD hadn't released Ryzen.

And by the time that 6core Intel CPU will be released, Zen 2 will be released also.

6core Intel will have to fight Zen 2.

[CruNcher]

There would be no 6core CPU for mainstream desktop from Intel, if AMD hadn't released Ryzen.

Of curse that's called competition no one is making you a present of more performance lower consumption for the same or lower cost it all comes through competition :)

And yes overall Ryzen is nice as i said best AMD processor since Athlon 64 after that i left AMD for Intel same as i left ATI after the Radeon 9000 series for Nvidia ;)

and in both cases now AMD gave partly up on their own paths and adapting it's Technology and gains traction again in both fields chasing it's competition now based on mostly the same Cores with still slightly different approaches though which both combine into 1 unique Ecosystem which Alpha test we saw with the Consoles starting ;)

[Atak_Snajpera]

Leaked Ryzen 5 results
https://www.purepc.pl/image/news/2017/03/24_procesory_amd_ryzen_5_wyciekly_do_sprzedazy_za_granica_2.jpg

It shows that Ryzen in SSE2 is only slightly faster than Sandy/Ivy Bridge. I see similar performance in flops. Coincidence? I do not think so.

[Sagittaire]

Leaked Ryzen 5 results

It shows that Ryzen in SSE2 is only slightly faster than Sandy/Ivy Bridge. I see similar performance in flops. Coincidence? I do not think so.

absoluty not, one more time: bad information or alternative fact ... :devil:

http://www.guru3d.com/index.php?ct=articles&action=file&id=28906&admin=0a8fcaad6b03da6a6895d1ada2e171002a287bc1

On single thread R7 1800X is on par with i7-7700K. If you make benchmark in score/frequency you have:
i7-7700K: 510 unit/ghz
R7 1800X: 540 unit/ghz

On single thread, R5 1600 with turbo at 3.6 Ghz (single thread) produce the same and coherent result in your CPUz benchmark:
R5 1600: 548 unit/ghz

http://www.guru3d.com/index.php?ct=articles&action=file&id=28961&admin=0a8fcaad6b03da6a6895d1ada2e171002a287bc1


For CPUz benchmark in MT, Rysen is simply a Intel Killer: R5 1600 6C/12T produce simply better result than i7-6900K 8C/16T here:

http://www.guru3d.com/index.php?ct=articles&action=file&id=28907&admin=0a8fcaad6b03da6a6895d1ada2e171002a287bc1


source: Guru3d (http://www.guru3d.com/articles-pages/amd-ryzen-7-1800x-processor-review,9.html)

[Atak_Snajpera]

I don't care about cpu-z. This bench is a mistery. We know zero about nature of calculations. How many integer and how many float tests are there? How does it calculate those scores? Does it use SIMD and so on...

Look at cinebench (SSE2).

[CruNcher]

absoluty not, one more time: bad information or alternative fact ... :devil:

http://www.guru3d.com/index.php?ct=articles&action=file&id=28906&admin=0a8fcaad6b03da6a6895d1ada2e171002a287bc1

On single thread R7 1800X is on par with i7-7700K. If you make benchmark in score/frequency you have:
i7-7700K: 510 unit/ghz
R7 1800X: 540 unit/ghz

On single thread, R5 1600 with turbo at 3.6 Ghz (single thread) produce the same and coherent result in your CPUz benchmark:
R5 1600: 548 unit/ghz

http://www.guru3d.com/index.php?ct=articles&action=file&id=28961&admin=0a8fcaad6b03da6a6895d1ada2e171002a287bc1


For CPUz benchmark in MT, Rysen is simply a Intel Killer: R5 1600 6C/12T produce simply better result than i7-6900K 8C/16T here:

http://www.guru3d.com/index.php?ct=articles&action=file&id=28907&admin=0a8fcaad6b03da6a6895d1ada2e171002a287bc1


source: Guru3d (http://www.guru3d.com/articles-pages/amd-ryzen-7-1800x-processor-review,9.html)

Keeping power requirements for Workloads under the hood is very unimpresive and Frequency out of the sweetspot causes massive power.

And only the fewest Reviewer can measure power requirements in some way efficient guru3d is not one of them.

[Sagittaire]

I don't care about cpu-z. This bench is a mistery. We know zero about nature of calculations. How many integer and how many float tests are there? How does it calculate those scores? Does it use SIMD and so on...

Look at cinebench (SSE2).

Well Rysen produce certainely best result in this bench: AMD use cinebench for these technical demonstration ... :D

http://www.guru3d.com/articles_pages/amd_ryzen_7_1800x_processor_review,10.html

One more time: bad information ... :devil:

[NikosD]

Cinebench is probably the best SW of ALL out there to demonstrate RyZen's multithreaded performance.

It's like RyZen was built for Cinebench, the favorite benchmark of Intel in previous years (before RyZen's release)

It should be a shock for Intel fan boys to loose so badly in their favorite benchmark.

[Sagittaire]

Cinebench is probably the best SW of ALL out there to demonstrate RyZen's multithreaded performance.

It's like RyZen was built for Cinebench, the favorite benchmark of Intel in previous years (before RyZen's release)

It should be a shock for Intel fan boys to loose so badly in their favorite benchmark.

yes ... and Atak_Snajpera show simply that little R5 1600 will be better than i7-7700K in this bench

[Atak_Snajpera]

Well Rysen produce certainely best result in this bench: AMD use cinebench for these technical demonstration ... :D

http://www.guru3d.com/articles_pages/amd_ryzen_7_1800x_processor_review,10.html

One more time: bad information ... :devil:

What is your problem? Can't you do simple math?

Ryzen 1700@4GHz = 1596 CB
https://youtu.be/Dan36jGqmzE?t=8m35s

1596 * 0.75 = 1197 cb for Ryzen 5 (6C/12T) @ 4GHz

[CruNcher]

Amd does this because Cinebench is Intel optimized from bottom to ground ;)

https://www.youtube.com/watch?v=PzLxCo5qofo

And not because that is so bad but the entire difference since Ryzen even unoptimized ;)

also these tests are mostly flawwed because Ryzen can never holdup 4 Ghz under full Core Load

http://i.cubeupload.com/vCYY13.jpg

[Sagittaire]

What is your problem? Can't you do simple math?

Ryzen 1700@4GHz = 1596 CB
https://youtu.be/Dan36jGqmzE?t=8m35s

1596 * 0.75 = 1197 cb for Ryzen 5 (6C/12T) @ 4GHz

yes ... but your initial affirmation is:

It shows that Ryzen in SSE2 is only slightly faster than Sandy/Ivy Bridge. I see similar performance in flops. Coincidence? I do not think so.

... and it's completely false ... :devil:

Rysen is just slightly faster than Broadwell-E in cinebench ... ;-)

http://www.guru3d.com/index.php?ct=articles&action=file&id=28968&admin=0a8fcaad6b03da6a6895d1ada2e171002a287bc1

http://www.guru3d.com/index.php?ct=articles&action=file&id=28969&admin=0a8fcaad6b03da6a6895d1ada2e171002a287bc1

[CruNcher]

It is crazy that wee see reviewers only measuring Encoding thereby Decoding is much more time critical and overall more stressful for the System in terms of overall stability ;)

And such 8 cores 16 threads perfect for 4K (UHD) High Bitrate Decoding efficiency tests also of instructions and overall optimization and it can be much better correlated where problems occur in near realtime by just listening to the Fan behavior and it's overall clock switching efficiency on the OS side and scheduling made hearable under it's full Power output.

you stressing a real scenario also on the Driver side :)

http://i1.sendpic.org/i/kb/kbOlJXnoiYyTezTj0OXLKzykIhp.png

[Atak_Snajpera]

yes ... but your initial affirmation is:

Quote:
It shows that Ryzen in SSE2 is only slightly faster than Sandy/Ivy Bridge. I see similar performance in flops. Coincidence? I do not think so.
... and it's completely false ...

Problem Or you mad bro?
Knowing that Ryzen 5 (6C/12T) @ 4GHz reaches 1200 CB we can calculate average clock frequency during test.
1100 CB / 1200 CB * 4GHz = 3.7 GHz
Max turbo for 3930k is 3.8 GHz

http://i.imgsafe.org/6c3dcbe5a4.png
http://i.imgsafe.org/6c4a9ed938.png

[CruNcher]

It interesting to see how the price tumbled down and slightly up again :D

http://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i7-7700K+%40+4.20GHz&id=2874

http://www.cpubenchmark.net/compare.php?cmp[]=2874&cmp[]=2970&cmp[]=2905


hehe but thats even more funny

http://www.cpubenchmark.net/cpu.php?cpu=AMD+Ryzen+7+1700&id=2970

up down up down


most probably the SMT News :D


I do believe that Skylake levels of performance are out of reach for Zen, and Intel will have their new CPU coming out around the same time as Zen, so Zen will be two generations behind in performance compared to Intel - which is a far cry better than the six generations they are currently behind.

Zen+'s promised 15% boost, though, will best Skylake, so if they get that out in a year's time after Zen, then they may practically catch up to Intel.

He wasn't really right which is good for everyone it's only 1 Generation before time to market now for AVG user starting when Coffee Lake Hits :)

And 1 Generation is not really that much for a CPU when the Price for the competition is acceptable :)

it looks very sane on the CPU side now though very insane on the GPU with Nvidia getting each generation more crazy and sooner or late falling on it's nose ;)


Though i really would like to see some more tests on the Decoding side of Ryzen that CCX interconect latency issues could become more problematic here :)


https://community.amd.com/community/gaming/blog/2017/03/14/tips-for-building-a-better-amd-ryzen-system


WOW you rarely see that Disclaimer

2. AMD processors, including chipsets, CPUs, APUs and GPUs (collectively and individually "AMD processor"), are intended to be operated only within their associated specifications and factory settings. Operating your AMD processor outside of official AMD specifications or outside of factory settings, including but not limited to the conducting of overclocking using the Ryzen Master overclocking software, may damage your processor, affect the operation of your processor or the security features therein and/or lead to other problems, including but not limited to damage to your system components (including your motherboard and components thereon (e.g., memory)), system instabilities (e.g., data loss and corrupted images), reduction in system performance, shortened processor, system component and/or system life, and in extreme cases, total system failure. It is recommended that you save any important data before using the tool. AMD does not provide support or service for issues or damages related to use of an AMD processor outside of official AMD specifications or outside of factory settings. You may also not receive support or service from your board or system manufacturer. Please make sure you have saved all important data before using this overclocking software.

They explain to you that Netflix could stop working ;)

What is uber funny is how they tell users hey overclock your memory and you get better Render results ehhhh note to AMD you would see the same on Intel ;)

https://www.youtube.com/watch?v=qksXthUcbiQ
https://www.youtube.com/watch?v=TId-OrXWuOE
https://www.youtube.com/watch?v=HZPr-gNWdvI

[Sagittaire]

Problem Or you mad bro?
Knowing that Ryzen 5 (6C/12T) @ 4GHz reaches 1200 CB we can calculate average clock frequency during test.
1100 CB / 1200 CB * 4GHz = 3.7 GHz
Max turbo for 3930k is 3.8 GHz


well it's enough. Cinebench is certainely the best test in area to show Ryzen power.

Here the unit/ghz for cinebench R15 in single thread test:
Core i7-7700K: 42.44 (Kaby Lake)
Core i7-6700K: 42.14 (Kaby Lake)
Core i7-6900K: 41.62 (Broadwell E)
Core i7-4790K: 39.31 (Haswell)
Rysen 7 1800: 39.00
Core i7-5960X: 38.85 (Haswell-E)
Core i7-2600K: 32.63 (Sandy Bridge)


Here the unit/ghz for cinebench R15 in multi thread test:
Rysen 7 1800: 425.78
Core i7-6900K: 418.64 (Broadwell E)
Core i7-5960X: 408.75 (Haswell-E)
Rysen 5 1600: 323.52
Core i7-7700K: 218.18 (Kaby Lake)
Core i7-6700K: 219.75 (Kaby Lake)
Core i7-4790K: 209.00 (Haswell)
Core i7-2600K: 176.17 (Sandy Bridge)


Here the unit/ghz/core for cinebench R15 in multi thread test:
Core i7-7700K: 54.54 (Kaby Lake)
Core i7-6700K: 54.93 (Kaby Lake)
Rysen 5 1600: 53.92
Rysen 7 1800: 53.22
Core i7-6900K: 52.33 (Broadwell E)
Core i7-4790K: 52.25 (Haswell)
Core i7-5960X: 51.09 (Haswell-E)
Core i7-2600K: 44.04 (Sandy Bridge)


You understand now that Rysen 7 is by far more performant for cinebench R15 than Sandy Bridge or Ivy Bridge:
- Ryzen is on par with Haswell for single thread
- Ryzen is on par with Broadwell E for Multi thread

You can note that SMT for Rysen is more efficient than Hyperthreading for cinebench.

It's clear now?

[NikosD]

I have updated my x264 & x265 benchmark results of this thread here:
https://forum.doom9.org/showthread.php?p=1800781#post1800781

Take a look.
It seems that RyZen has better IPC for x265 than x264 using the default settings of the benchmark, compared to older Intel architectures (Sandybridge & Haswell) but not for Skylake.
I have updated again my results with accurate RyZen tests and with a huge surprise

Ryzen running on 1+1 is faster than 2+0 (!)

Read more on my above post.

[Sagittaire]

and we are on doom9 and result is even better (but coherent) for x264:

http://www.hardware.fr/getgraphimg.php?id=465&n=9

it's clear now?

[Sagittaire]

I have updated again my results with accurate RyZen tests and with a huge surprise

Ryzen running on 1+1 is faster than 2+0 (!)

Read more on my above post.

yes ... because you have really higher cache in 1+1 than 2+0 mode. In fact twice for 1+1 vs 2+0.

and be carefull, SMT for ryzen in more efficient than Hyperthreading. Certainely that 2C/4T test will be even better for Ryzen.

http://www.hardware.fr/getgraphimg.php?id=437&n=1

[Atak_Snajpera]

You do realize that x264 mainly rely on performance of integer units in cpu,right? Cinebench is different. Rendering almost exclusively runs on FPU. For example Cinebench only uses SSE2 SIMD.
Clock vs Clock SSE2 unit in RyZen is only slightly faster than in SandyBridge-E. Deal with it.

Now I see that you are an another corporate fanboy. Numbers below clearly show that RyZen = SandyBridge-E in floating-point calculations.
http://i.imgsafe.org/6c3dcbe5a4.png

and be carefull, SMT for ryzen in more efficient than Hyperthreading. Certainely that 2C/4T test will be even better for Ryzen.
It is more efficient because it starts from lower level!
http://i.imgsafe.org/79a616c1ff.png

[Sagittaire]

You do realize that x264 mainly rely on performance of integer units in cpu,right? Cinebench is different. Rendering almost exclusively runs on FPU. For example Cinebench only uses SSE2 SIMD.
Clock vs Clock SSE2 unit in RyZen is only slightly faster than in SandyBridge-E. Deal with it.

Now I see that you are an another corporate fanboy. Numbers below clearly show that RyZen = SandyBridge-E in floating-point calculations.


No ... you are fanboy corporate ...

I post clearly all the result that you can find in all the preview: and conclusion is really good for Rysen in cinebench. Cinebench R15 is the test that AMD itself use for make technical demonstration for Rysen.

I don't know were you find your result. But I post the result that you can find in all independant preview:

Here the unit/ghz for cinebench R15 in single thread test:
Core i7-7700K: 42.44 (Kaby Lake)
Core i7-6700K: 42.14 (Kaby Lake)
Core i7-6900K: 41.62 (Broadwell E)
Core i7-4790K: 39.31 (Haswell)
Rysen 7 1800: 39.00
Core i7-5960X: 38.85 (Haswell-E)
Core i7-2600K: 32.63 (Sandy Bridge)


Here the unit/ghz for cinebench R15 in multi thread test:
Rysen 7 1800: 425.78
Core i7-6900K: 418.64 (Broadwell E)
Core i7-5960X: 408.75 (Haswell-E)
Rysen 5 1600: 323.52
Core i7-7700K: 218.18 (Kaby Lake)
Core i7-6700K: 219.75 (Kaby Lake)
Core i7-4790K: 209.00 (Haswell)
Core i7-2600K: 176.17 (Sandy Bridge)


Here the unit/ghz/core for cinebench R15 in multi thread test:
Core i7-7700K: 54.54 (Kaby Lake)
Core i7-6700K: 54.93 (Kaby Lake)
Rysen 5 1600: 53.92
Rysen 7 1800: 53.22
Core i7-6900K: 52.33 (Broadwell E)
Core i7-4790K: 52.25 (Haswell)
Core i7-5960X: 51.09 (Haswell-E)
Core i7-2600K: 44.04 (Sandy Bridge)

It's clear that you make trolling now ...

[Atak_Snajpera]

I don't know were you find your result. But I post the result that you can find in all independant preview:
Results are hardcoded in Cinebench app you dumb ass! Have you ever run cinebench r15 on your PC?!?
http://i.imgsafe.org/79d8a4a197.png

[CruNcher]

I have updated again my results with accurate RyZen tests and with a huge surprise

Ryzen running on 1+1 is faster than 2+0 (!)

Read more on my above post.

What is the overall result for 3+3 ?

Results are hardcoded in Cinebench app you dumb ass! Have you ever run cinebench r15 on your PC?!?
http://i.imgsafe.org/79d8a4a197.png

please dont forget they will slightly differ by Windows OS


Passmarks testing is also overall much more reliable imho and represents very accurately CPU Performance and you have access to the entire database of test and they have a sample test algorithm with error margin

[NikosD]

You mean 4+4 and it's in the end of my post above.

[CruNcher]

no i mean 3+3 actually preview what Ryzen 5 will bring on the table :)


6 cores will be the new Mainstream target 6/12

8 cores full utilization will take some time for every Software Developer to achieve properly Low Level and get good scaling results mostly driven by Scorpio ;)

if i had the decisions to efficiently upgrade i would rather put my money into 6 Cores and save the rest for the GPU side instead of 2 additional cores which performance delta isn't optimal yet due to it's interconnect and also save energy cost until Software is fully ready and most probably get the internal IGPU for free too ;)

Why should i just go for 2 cores if i can use the internal IGPU to lower Workload pressure by much more then the 2 cores can help higher Performance, that would be dumb ;)

[Sagittaire]

Results are hardcoded in Cinebench app you dumb ass! Have you ever run cinebench r15 on your PC?!?


No, Cinebench don't indicate the real frequency, just stock frequency. R5 1600 is not disponible for official test and you don't know how work really turbo: certainely 3.2/3.4/3.6 for Stock/Turbo max all core/turbo max one core. Moreover You don't know what are the real frequency for i7-3930K: stock is 3.2 Ghz and max turbo for one core is 3.8 Ghz and reall frequency for cinebench MT will be certainely between 3.2 and 3.8 Ghz but you can't say more.

I post independant result from Guru3d and I can post many other result from Ryzen preview if you want. And in all these preview, Ryzen is an Intel Killer: A simple R7 1700 at 329$ will be on par with i7-5960X or 17-6900K at 1000$.

http://www.anandtech.com/show/11170/the-amd-zen-and-ryzen-7-review-a-deep-dive-on-1800x-1700x-and-1700/18

http://images.anandtech.com/graphs/graph11170/85881.png

[CruNcher]

I wonder how surprised you will be Sagittaire if you see Intels 6 Cores beating that 8 Cores end of 2017 and already beating it in the Lab ;)

My i5 2400 btw is in that range of the i3 Kaby Lake by now

http://www.cpubenchmark.net/compare.php?cmp[]=2930&cmp[]=793&cmp[]=2970

But yeah you cant really deny Ryzen 7 1700 is hell of a good deal and Future investment with that Scorpio Backup coming :)

And that Glimpse shows the potential, Eidos Labs and Nixxes already do a pretty good job with the ground up work on the Dawn Engine

http://media.bestofmicro.com/ext/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS8wLzQvNjU5NjY4L29yaWdpbmFsL2ltYWdlMDUyLnBuZw==/r_600x450.png

[Sagittaire]

I wonder how surprised you will be Sagittaire if you see Intels 6 Cores beating that 8 Cores end of 2017 and already beating it in the Lab ;)

My i5 2400 btw is in that range of the i3 Kaby Lake by now

perhaps ... but it's not actually the case ... ;-)

Actually R7 1800X at 4.0 Ghz produce twice more performance than i7-6700K at 4.0 ghz in cinebench R15 MT.

If you want make that with 6C/12T Kaby Lake, you must have something like 5.0 Ghz for stock frequency. I seriousely doubt that Intel make that in 2017.
i7-7700K have actually low power efficiency and Intel is certainely at max clock for Kaby Lake for i7-7700K for correct power efficiency.

Power Efficacity (100*fps/watt) for x264 Encoding (higher is better)
http://www.hardware.fr/getgraphimg.php?id=464&n=1

[CruNcher]

Yeah actually i wonder also how they could achieve it especially with the IGPU overhead on practical the same node but it's Intel ;)

If they don't make it AMD looks really in the lead at least without IGPU overhead.


Raven Ridge is becoming more and more exciting to look out for for Laptops the more Zen data spreads vs i7-7700HQ :D

http://www.cpubenchmark.net/compare.php?cmp[]=2906&cmp[]=2937&cmp[]=2970


Did somebody do a 2+2 test yet and looked at the power consumption ?

though if you just calculate with 2 you roughly get @ 7000 not enough to beat Intel Mobile on the CPU side of course it would be pretty enough to beat Intel on the complete Platform with Vega inside this time no doubt about that even with slightly lower IPC Intel would see no sunlight this time (with their shrinking advantage made pooof) in most Workloads competitive and in 3D unbeatable.

Jensen surely doesn't sleeps well currently as well hoping to sell some GTX 1050 bricks fast ;)

[Sagittaire]

Yeah actually i wonder also how they could achieve it especially with the IGPU overhead on practical the same node but it's Intel ;)

If they don't make it AMD looks really in the lead at least without IGPU overhead.


Raven Ridge is becoming more and more exciting to look out for for Laptops the more Zen data spreads :D

be carefull in these previews, IGPU for intel is in off mode (configuration use GTX 1080 in most case). Moreover in idle mode, iGPU certainely don't use more than 5 watts and will not change efficiency result for CPU.

[CruNcher]

I think there is a big possibility that Raven Ridge is gonna land here

http://www.cpubenchmark.net/cpu.php?cpu=Intel+Core+i3-7350K+%40+4.20GHz&id=2930

but with much much better GPU Performance of course :)

Also Sagittaire that 5 GHz are reachable for Intel already it surely takes not so much of fine tuning and optimization to reach it stable for the 6 Coffe Lake Cores surely you cant expect them to reach 5 GHz with full threading though

https://youtu.be/dxgrb0YR_fA?t=309

seeing such experimental results means they working hard in the lab to tighten them ;)

i wouldn't be surprised if they would dethrone Raven Ridge again 2018 with that 6 Core Mobile im actually pretty sure they will this is how the cycle works if you 1 Generation behind.

But at least the Prediction of AMD 2 Generations behind i wouldn't share it's 1 Generation 1 Cycle pretty perfect for staying alive and share the market :)


Richards next FCAT Compares are out looking at the imho super nice Ryzen 7 1700

https://www.youtube.com/watch?v=-RRt5WkVxuk

AMD made a good decision putting two on top that have slightly higher performance but aren't worth it to catch zillion of more buyers for the Ryzen 7 1700 that is a great Strategy that's gonna succeed :)

everyone will see the Value of the Ryzen 7 1700 and understand how fair it is


Especially how adaptive you can use that system and save even more resources not going all the time with the full 8 cores but switching between 6 and 8 cores depending on the Workload for now until the 8 cores get throughout used ;)


i would actually pretty much disable 2 cores and save the energy and use only the 6 cores at first for most Workloads and use those 2 extra cores only as backup till they widely low level get used on the consumer side of things ;)

https://youtu.be/-RRt5WkVxuk?t=592

He should know better that Intel is already preparing the answer and it's running in the Lab and might be even sooner released then first anticipated, intel will look closely at its market share tumbling now and might push a lot of money and scale their schedule based on that for a pretty fast release ;)

It's funny to see Reviewers that don't understand the Game they Playing in ;)

also these Video Encoding compares on the Software side with those archiving targets i find them economically blatantly idiotic for avg consumer and even for semi professional content producer but that would be worth another topic.

This is also a reason i find it pretty unfair how the R&D inside of Kaby Lake and Balancing decision gets judged by most "Pro Reviewers" overall it's so blatantly wrong and it gets pretty easy to sell those Gamer target something which is technically very questionable overall even if others want to make you believe the investment is worth it with so blatantly wrong arguments ;)

Especially if those Reviewer use Workloads from the last 20 century and don't understand Platform Balance overall.

[Sagittaire]

Also Sagittaire that 5 GHz are reachable for Intel allready it surely takes not so much of fine tuning and optimization to reach it stable for the 6 Coffe Lake Cores surely you cant expect them to reach 5 GHz with full threading though

Certainely for extreme overclocking but not for stock frequency.

- If you want make that with 6C/12T Kaby Lake at 5 Ghz stock, you have certainely something like 160-180 watt power consumption (and perhaps higher).
- If Intel make 6C/12T Kaby Lake at 5 Ghz stock, then the power efficacity will be really low, really lower than Sandy Bridge at stock frequency (i7-2600K and i7-7700K are on par for x264 power efficiency)
- For each new architecture, Intel annonce generaly 10% more IPC and not more.

I don't think that Intel can make quickly a new 6C/12T CPU better than R7 1800X. Anyway Intel can make certainely really quickly a 8C/16T Kaby Lake CPU (i7-7900K at 4.0/4.2/4.5 with TDP at 140 watts for exemple) certainely really faster than R7 1800X in all case.

[NikosD]

Intel can't do anything right now.

They can only sit down and watch their new HEDT Skylake X platform getting destroyed by AMD 16C/32T HEDT CPU and that 6C/12T is going to face Zen 2.

So...

[Sagittaire]

They can only sit down and watch their new HEDT Skylake X platform getting destroyed by AMD 16C/32T HEDT CPU and that 6C/12T is going to face Zen 2.


there are rumor for R7 1950X 16C/32T on desktop at 1000$ ... :D

samples circulate for validation in professional area

[CruNcher]

It's a Dual Socket System rarely any AVG Consumer Target though some Gamers are surely crazy even if most Numa nodes would idle all the time and consuming power for nothing some Gamers would surely buy those believing they get something from it ;)

i have to say i understand more Gamers that buy overpriced laptops then these Systems for Gaming purposes hell even would tell them buy some all color glowing Razer shit instead and be happier seeing that colors flow instead of seeing your nodes idle :D

Which we though currently see very often with Ryzen as well, according to youtube data ;)

Even with Semi Pro targeted Video Editing you will still experience that ;)