GPU doesn't affect scores. I don't think memory does either, but I can't say for sure.
But your results show how well Cinebench scales with core count as opposed to clock speed. It is far ahead of a 7700K, which has a nearly 1GHz higher clock but two less physical cores.
Except you're wrong because the 6900K has 8 cores and scores just slightly more than Ryzen 1700X which is inline with every other benchmark. So it's not really 'far ahead' given the core count and it's position as HEDT at the time of release.
You'll always see decent scaling with frequency unless there is a bottleneck feeding the instructions or whatever. "Cinebench scales with core count as opposed to clock speed" makes no sense whatsoever. What you meant to say that it scales with core count well. Scaling with frequency is implicit.
The only hexa core that would score >3500 is the 8700K anyway, just use common sense.
3930K @ 3.2ghz, 16GB DDR3 @ 1600mhz, Radeon RX 580. - 2020 points. Actually surprised how near this 8+ year old rig gets to more modern machines at stock clocks.
I wanna try it at 4.8Ghz now to see if I can beat that Ryzen... Damn Australian summer.
About the same performance per-core! Of course if AES, AVX or SSE4.1/4.2 came into play it might be another story. With everything embracing multithreading, though, it's probably aged quite well.
i7-8550u (Dell Latitude 7390 2-in-1): 1229 - for five consecutive runs, no less.
The lack of variance in the scores is kind of frightening. This was across battery and AC powered runs, and at stock settings as well as -80mV undervolt core/cache and PL1/PL2 set to 25/40W (from stock 18/25W). I don't think XTU actually works on this PC (outside of any undervolt causing BSODs on sleep, likely due to "modern standby"), but besides that, getting identical scores across five consecutive benchmark runs is _weird_. Either this is the most consistent benchmark ever, or this PC has ... OCD or something.
It's worth mentioning that this laptop has the smallest heatsink I've ever seen, and yet it stays impressively cool under load (usually in the high 70s or low 80s) without much noise. XTU reports sustained power draw around 12-13W.
Wow this is actually awesome someone is doing this. If they had a "DVR" mode that could really simplify scaling security camera installations without unnecessarily overbuilding.
My current rule of thumb is at least a Core i7-8700 per 10 4K (8 megapixel) video feeds and that's using Quicksync as a encoding aid. The problem is a lot of software (such as Sighthound) doesn't use Quicksync so I have it take the 2nd feed (h265 720p) from security cameras for analysis and have it dump the 1st feed (h265 4k) direct to disk.
Blue Iris supports Quicksync so it can encode a h264 4k stream, around 10 at a time, on a Core i7-8700.
The problem is this is all subjective and often overbuilt. And when dealing with, say, 45 cameras in a multi-warehouse installation there is no 'single CPU' solution so I have to build at least two DVR's and segment them. It'd be really nice to just plug in what my recording\encoding requirements are and have a benchmark app tell me what CPU is needed.
Download Cinebench R20 from Windows store Launch App Open Task manager, Details tab, find Cinebench.exe, Right click, select Open File Location This will open a “bin” folder Copy entire bin folder You now have a standalone copy of R20
Makes me wish I still had my old C2Q Q9450 (@3.52GHz). Would have been interesting to see how it kept up (or not). Still, pretty impressive that my 15W (well, 12W according to XTU) i7-8550u beats that 95W CPU by nearly 2.2x.
I have Xeon e5450. I can check score in R20 this afternoon. I run it normally at 3870 MHz. If you want, I can run it at 3.52Ghz, than it should be equal to your old CPU. Confirm if you want lower freq, otherwise I will post score with my clock later. It should scale linear tho even with q6600, so it should be about 750 pts for 3.52Ghz and 828 for 3.87Ghz.
9700k, stock clocks (8C no HT), 16GB @ 3200Mhz, 3543 score. It's good to see that a newer mainstream i7 has almost caught up with the 6900k, a two year old 8C16T enthusiast grade CPU which cost almost three times as much at the time.
I run Windows 2019 server, basically the same as Windows 10 1809, except that it doesn't support the Windows store and thus I cannot download the benchmark.
Now I actually consider that a feature, because I cannot think of any benefit the Microsoft store would have to consumers. But it seems Microsoft would rather like to force the planet now to pay a Microsoft tax on all software that runs on their OS or just uses some of their APIs (WINE beware).
Considering that Microsoft has been properly punished for including their media players and browsers into an operating system, this approach seems a major violation of market power which deserves full anti-trust attention and swift fines at a level impossible not to learn from..
Server 2019 is also missing most of the snooping features of desktop windows variants, perhaps the ony Microsoft OS actually capable to operate under GDPR compliance. Another area where privacy authorities should urgently exert serious authority and steer Microsoft away from behavior that should cost it dearly to pursue.
5196 on E5-2696v3 after I found out how to sideload it without a store, CPU never pushing beyond 58°C and 98 Watts according to HWinfo 6.02, 2.6GHz on all 18 cores and 36 threads.
i just got 289 with my shit old hp laptop amd a4-6210apu with r3 graphics 4 cores @1.8 ghz and 8 gb ram . i looking at building a pc plus i am getting a old i5 2400 to play with i just wanted to see how slow this was as a base and will see what these other pc will do when i can
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
57 Comments
Back to Article
Sttm - Tuesday, March 5, 2019 - link
6900k stock clocks, 32gb at @2667, GTX1080 and I get a 3660.jordanclock - Tuesday, March 5, 2019 - link
GPU doesn't affect scores. I don't think memory does either, but I can't say for sure.But your results show how well Cinebench scales with core count as opposed to clock speed. It is far ahead of a 7700K, which has a nearly 1GHz higher clock but two less physical cores.
Yesumanu - Wednesday, March 6, 2019 - link
Except you're wrong because the 6900K has 8 cores and scores just slightly more than Ryzen 1700X which is inline with every other benchmark. So it's not really 'far ahead' given the core count and it's position as HEDT at the time of release.You'll always see decent scaling with frequency unless there is a bottleneck feeding the instructions or whatever. "Cinebench scales with core count as opposed to clock speed" makes no sense whatsoever. What you meant to say that it scales with core count well. Scaling with frequency is implicit.
The only hexa core that would score >3500 is the 8700K anyway, just use common sense.
silverblue - Tuesday, March 5, 2019 - link
Ryzen 5 1600 stock (though it seemed locked at 3.4GHz throughout), 16GB (2x8) @ 3000MHz - 2542 points, so it does just overhaul the 7700K.342 single thread, core clock 3.4 to 3.45GHz - definitely not as high as I'd expect given a max. clock speed of 3.7GHz with XFR.
StevoLincolnite - Wednesday, March 6, 2019 - link
3930K @ 3.2ghz, 16GB DDR3 @ 1600mhz, Radeon RX 580. - 2020 points.Actually surprised how near this 8+ year old rig gets to more modern machines at stock clocks.
I wanna try it at 4.8Ghz now to see if I can beat that Ryzen... Damn Australian summer.
FatalError - Tuesday, March 5, 2019 - link
a paltry 509 points on a Surface Book with performance base @i7-6600UGreenReaper - Tuesday, March 5, 2019 - link
AMD [email protected] dual-core (Brazos/Bobcat), 6GB in a Lenovo x120e - 57 pointsSmell This - Wednesday, March 6, 2019 - link
AMD A6-1450 @ 1.4GHz 8W quad-core Temash -- 106cbHP Pavilion TouchSmart 11, Win8 64-bit, 8GB, SS 120GB 840 Evo
(still 'smoking')
GreenReaper - Thursday, March 7, 2019 - link
About the same performance per-core! Of course if AES, AVX or SSE4.1/4.2 came into play it might be another story. With everything embracing multithreading, though, it's probably aged quite well.maoparmannz - Tuesday, April 2, 2019 - link
289 AMD A4 6210 . 4 CORE @1.8Ashinjuka - Tuesday, March 5, 2019 - link
4770K stock, 16GB @ 1600MHz: 1668sbrown23 - Wednesday, March 6, 2019 - link
i7-3770K at 4.1GHz turbo (3.8 all cores), 32GB RAMMulti - 1379
Single - 297
sbrown23 - Wednesday, March 6, 2019 - link
i7-4771 at stock 3.5GHz in iMac 27" Late-2013 model (macOS 10.14.3) with 24GB RAMMulti - 1655
Single - 331
CoachAub - Tuesday, March 5, 2019 - link
This is great! Ryzen 7 1800X (stock clocks), 16gb @3000 w/RX 480 gpu. Score=354029a - Tuesday, March 5, 2019 - link
Ryzen 7 2700X 16 GB 3200 16-18-18 1T Stock 3862 PBO 393669369369 - Tuesday, March 5, 2019 - link
i5-8250U with -100mV undervolt to core and cache1312 (on battery, 29W/15W)
1594 (plugged in, 44W/25W)
Valantar - Wednesday, March 6, 2019 - link
i7-8550u (Dell Latitude 7390 2-in-1): 1229 - for five consecutive runs, no less.The lack of variance in the scores is kind of frightening. This was across battery and AC powered runs, and at stock settings as well as -80mV undervolt core/cache and PL1/PL2 set to 25/40W (from stock 18/25W). I don't think XTU actually works on this PC (outside of any undervolt causing BSODs on sleep, likely due to "modern standby"), but besides that, getting identical scores across five consecutive benchmark runs is _weird_. Either this is the most consistent benchmark ever, or this PC has ... OCD or something.
Valantar - Wednesday, March 6, 2019 - link
It's worth mentioning that this laptop has the smallest heatsink I've ever seen, and yet it stays impressively cool under load (usually in the high 70s or low 80s) without much noise. XTU reports sustained power draw around 12-13W.BigMamaInHouse - Wednesday, March 6, 2019 - link
TPU added download link:https://www.techpowerup.com/download/maxon-cineben...
Xidash - Friday, March 8, 2019 - link
Could you do single core score please ?Samus - Tuesday, March 5, 2019 - link
Wow this is actually awesome someone is doing this. If they had a "DVR" mode that could really simplify scaling security camera installations without unnecessarily overbuilding.My current rule of thumb is at least a Core i7-8700 per 10 4K (8 megapixel) video feeds and that's using Quicksync as a encoding aid. The problem is a lot of software (such as Sighthound) doesn't use Quicksync so I have it take the 2nd feed (h265 720p) from security cameras for analysis and have it dump the 1st feed (h265 4k) direct to disk.
Blue Iris supports Quicksync so it can encode a h264 4k stream, around 10 at a time, on a Core i7-8700.
The problem is this is all subjective and often overbuilt. And when dealing with, say, 45 cameras in a multi-warehouse installation there is no 'single CPU' solution so I have to build at least two DVR's and segment them. It'd be really nice to just plug in what my recording\encoding requirements are and have a benchmark app tell me what CPU is needed.
CleverBullet - Tuesday, March 5, 2019 - link
I think you replied to the wrong article, this if for Cinebench.austinsguitar - Wednesday, March 6, 2019 - link
xDlooncraz - Tuesday, March 5, 2019 - link
Apparently we have to get it through an app store??This trend needs to stop.
Alistair - Tuesday, March 5, 2019 - link
It's free. What better way to get an up to date free piece of software?Ian Cutress - Tuesday, March 5, 2019 - link
Download Cinebench R20 from Windows storeLaunch App
Open Task manager,
Details tab,
find Cinebench.exe,
Right click,
select Open File Location
This will open a “bin” folder
Copy entire bin folder
You now have a standalone copy of R20
shabby - Tuesday, March 5, 2019 - link
Windows has an app store? Whodathunkit...TestOfTime - Tuesday, March 5, 2019 - link
Coming to an AMD launch event near you.beaker7 - Tuesday, March 5, 2019 - link
13859 score on my email / web machine (dual xeon 6154)BigMamaInHouse - Tuesday, March 5, 2019 - link
Maybe Cinebench R15 Extreme edition push them to launch? :-)Ian Cutress - Wednesday, March 6, 2019 - link
Nah, this has been in the works a long while. Couldn't post about it, obviously.Epg - Tuesday, March 5, 2019 - link
Core 2 Quad Q6600 @2.4GHz, 8GB DDR2-800 : 514 pointsValantar - Wednesday, March 6, 2019 - link
Makes me wish I still had my old C2Q Q9450 (@3.52GHz). Would have been interesting to see how it kept up (or not). Still, pretty impressive that my 15W (well, 12W according to XTU) i7-8550u beats that 95W CPU by nearly 2.2x.adamo1139 - Wednesday, March 6, 2019 - link
I have Xeon e5450. I can check score in R20 this afternoon. I run it normally at 3870 MHz. If you want, I can run it at 3.52Ghz, than it should be equal to your old CPU. Confirm if you want lower freq, otherwise I will post score with my clock later. It should scale linear tho even with q6600, so it should be about 750 pts for 3.52Ghz and 828 for 3.87Ghz.bigvlada - Wednesday, March 6, 2019 - link
Xeon e5450 @3GHz, 8GB DDR2-800: 674 points.austinsguitar - Wednesday, March 6, 2019 - link
its a smol worldadamo1139 - Wednesday, March 6, 2019 - link
E5450 @3.87Ghz exacly 900cbAditya781 - Sunday, March 17, 2019 - link
Athlon X4 740 @3.73GHz : 544 pointsShivansps - Tuesday, March 5, 2019 - link
You should run this on AVX-512 cpus ASAP to find out if it supports AVX-512.Jorgp2 - Tuesday, March 5, 2019 - link
Still has the option to sort by GPU vendorLarsBolender - Wednesday, March 6, 2019 - link
7820X@4,6GHz4.0GHz AVX2
3,4GHz AVX512
4x4GB 3200CL15 + Subtimings
-> 4792 points
randomhkkid - Wednesday, March 6, 2019 - link
2811 / 419Razer Blade Advanced 2018 i7-8750H
Interesting to see that my scores is quite a bit lower relative to a stock 8700k in CBr20 vs CBr15
airdjing - Wednesday, March 6, 2019 - link
Xeon E5-2696v4, 64GB : Multi 6721 Point / Single 321 PointTopher46 - Wednesday, March 6, 2019 - link
9700k, stock clocks (8C no HT), 16GB @ 3200Mhz, 3543 score. It's good to see that a newer mainstream i7 has almost caught up with the 6900k, a two year old 8C16T enthusiast grade CPU which cost almost three times as much at the time.colonelclaw - Wednesday, March 6, 2019 - link
10912 on my work machine. Probably divide that by 10 for my home machine!DaveyTech - Wednesday, March 6, 2019 - link
1292 points with a Phenom II x6 @ 3.9GhzMentawl - Wednesday, March 6, 2019 - link
8700k @ 4.7ghz all core = 480 single threaded, 3571 multi.Lord of the Bored - Wednesday, March 6, 2019 - link
486 DX4 @ 100 MHz. 12 pointsGreenReaper - Thursday, March 7, 2019 - link
Mmm . . . and how'd you make that work without SSE3 or even 64-bit support?Xidash - Thursday, March 7, 2019 - link
Ryzen 7 1700 @ 3.725 ghz all coresRam 16 gb 3200 mhz 14-14-14-34
No offset
Multi : 3724
Single : 379
abufrejoval - Saturday, March 9, 2019 - link
I run Windows 2019 server, basically the same as Windows 10 1809, except that it doesn't support the Windows store and thus I cannot download the benchmark.Now I actually consider that a feature, because I cannot think of any benefit the Microsoft store would have to consumers. But it seems Microsoft would rather like to force the planet now to pay a Microsoft tax on all software that runs on their OS or just uses some of their APIs (WINE beware).
Considering that Microsoft has been properly punished for including their media players and browsers into an operating system, this approach seems a major violation of market power which deserves full anti-trust attention and swift fines at a level impossible not to learn from..
Server 2019 is also missing most of the snooping features of desktop windows variants, perhaps the ony Microsoft OS actually capable to operate under GDPR compliance. Another area where privacy authorities should urgently exert serious authority and steer Microsoft away from behavior that should cost it dearly to pursue.
abufrejoval - Saturday, March 9, 2019 - link
5196 on E5-2696v3 after I found out how to sideload it without a store, CPU never pushing beyond 58°C and 98 Watts according to HWinfo 6.02, 2.6GHz on all 18 cores and 36 threads.Cinebench R15 on this CPU is 2552.
wbj - Thursday, March 28, 2019 - link
Direct download, here you go: https://www.maxon.net/en-us/support/downloads/zircosil - Saturday, March 9, 2019 - link
i5-4670K running @ 4.4GHz1559 points multi-threaded
407 points single core
maoparmannz - Tuesday, April 2, 2019 - link
i just got 289 with my shit old hp laptop amd a4-6210apu with r3 graphics 4 cores @1.8 ghz and 8 gb ram . i looking at building a pc plus i am getting a old i5 2400 to play with i just wanted to see how slow this was as a base and will see what these other pc will do when i canMAX_AB - Saturday, April 6, 2019 - link
Cinebench R20 Score 15,836 best Cinebench R15 is 6872System: Dual Xeon Supernmicro Workstation
CPU 2x Xeon Platinum 8168 es2 (QLKN), no OC.
Base 2.4GHz/Allcore 2.9GHz/Turbo 3.5GHz (0.3GHz lower than production 8168)
MBD Supermicro X11DPG-QT
GPU GTX 1080Ti
RAM 128GB Samsung DDR4 2400 C17 ECC
SSD WDC SN750 512GB NVME
PSU EVGA 1300G2
PDU Sceptre U27 UHD IPS 27"
Case Rosewill B2
Cooling- standard air, 2x Supermicro SNK00070APS
Settings: NUMA on, Onboard VGA on, EFI mode, default CPU settings
Gurthang77 - Thursday, July 18, 2019 - link
Intel i7 4790K (4 GHz) -- 1891