AMD Ryzen 7 1800X hits 5.2 GHz, breaks benchmarking record

[vorticalbox]

15 hours ago, Bananasplit_00 said:

im going to gues you can get 4.6GHz on water at the max and 4.3GHz on air.

which kinda sucks seeing as I can get 5ghz at 1.5 on my 8350.

[Mr.Meerkat]

15 hours ago, TrigrH said:

um? not really? the 7700k can reach that on water. Would be nice to see 4.5 out of an 1800x but thats looking pretty slim right now

Well the 7700K starts off at 4.5GHz stock (in the sense that 4.5GHz is the highest "stock" clock speed)...sooooo ya know...  

Also where are the 6/8 core Intel CPUs running at 5GHz? Yeah...exactly my point  

[MageTank]

15 minutes ago, vorticalbox said:

which kinda sucks seeing as I can get 5ghz at 1.5 on my 8350.

Your 5ghz 8350 would be absolutely destroyed by even a 3ghz Zen in basically every task ever. Never ignore the entire picture of IPC.

[TrigrH]

4 minutes ago, Mr.Meerkat said:

Also where are the 6/8 core Intel CPUs running at 5GHz? Yeah...exactly my point  

 

15 hours ago, TrigrH said:

Would be nice to see 4.5 out of an 1800x but thats looking pretty slim right now

5820k (Launch Date Q3'14) can get close to 4.8ghz (on a good chip) and 4.5 on all chips so I wouldn't be surprised if Intel can close the gap with their next enthusiast batch of CPUs?

[Mr.Meerkat]

3 minutes ago, TrigrH said:

 

5820k (Launch Date Q3'14) can get close to 4.8ghz (on a good chip) and 4.5 on all chips so I wouldn't be surprised if Intel can close the gap with their next enthusiast batch of CPUs?

I guess? However remember how broadwell-E CPUs actually overclocks worse than Hawell-E  

[Curufinwe_wins]

15 minutes ago, Mr.Meerkat said:

I guess? However remember how broadwell-E CPUs actually overclocks worse than Hawell-E  

For now....

 

Extremely early Haswell-E chips had huge issues reaching even 4.5 on most samples, but by about 6 months after launch, it was pretty much a gimme.

 

Same with early GM200 chips almost never exceeding 1520 MHz while later ones breezed past without issue.

 

Now that said, the same effect might be seen on Zen, but the point remains that I really wouldn't expect Zen to be capable of the same top end clocks as BW-E (and by the time it comes around, certainly not the same as Skylake/Kabylake -E/X, since that is one of the biggest sell points of the second and third generation 14nm processes).

[Mr.Meerkat]

2 minutes ago, Curufinwe_wins said:

Extremely early Haswell-E chips had huge issues reaching even 4.5 on most samples, but by about 6 months after launch, it was pretty much a gimme.

 

Same with early GM200 chips almost never exceeding 1520 MHz while later ones breezed past without issue.

 

Now that said, the same effect might be seen on Zen, but the point remains that I really wouldn't expect Zen to be capable of the same top end clocks as BW-E (and by the time it comes around, certainly not the same as Skylake/Kabylake -E/X, since that is one of the biggest sell points of the second and third generation 14nm processes).

Well who said this isn't the case for everything? 

Both the 290X and 390X had the same 'dies' but in general 390Xs were able to overclock higher (and consume slightly less at the same clocks), newer RX480 (wafers) are actually hitting AMD's original power consumption goals, Haswell-E chips as you've mentioned, Haswell to Haswell refresh.

They all have the same thing in common, the manufacturing technique/process is given time to mature where they are simply able to produce better wafers. 

 

First 14nm CPU chips for AMD, I'm sure we can give them a little slack if they don't overclock that well  

[vorticalbox]

43 minutes ago, MageTank said:

Your 5ghz 8350 would be absolutely destroyed by even a 3ghz Zen in basically every task ever. Never ignore the entire picture of IPC.

yeah I know still 5GHz is just a nice number LOL.

[Curufinwe_wins]

17 minutes ago, Mr.Meerkat said:

Well who said this isn't the case for everything? 

Both the 290X and 390X had the same 'dies' but in general 390Xs were able to overclock higher (and consume slightly less at the same clocks), newer RX480 (wafers) are actually hitting AMD's original power consumption goals, Haswell-E chips as you've mentioned, Haswell to Haswell refresh.

They all have the same thing in common, the manufacturing technique/process is given time to mature where they are simply able to produce better wafers. 

 

First 14nm CPU chips for AMD, I'm sure we can give them a little slack if they don't overclock that well  

I actually agreed it is mostly the same case for everything.

 

Little Hawaii Rant

Spoiler

But actually similar production time frame 290x and 390x, the 390x actually overclock marginally worse, but do do better with power consumption (marginally). This is 100% due to some changes in SOFTWARE power delivery, not anything else. But yea 390x to launch 290x is no contest.

 

That said, all Fiji gpus still seem to suck donkey dick at overclocking, and 14/16nm processes from non-intel seem to be hitting pretty rough class ceilings on frequency (even though 2100 MHz is no joke from Pascal, and 1400 or so is pretty darn good from Polaris, I have yet to see any significant improvements there.)

 

 

The real thing to realize though is the overclocking headroom to be gained as a process matures is 100% not about AMD, but about the companies manufacturing the chips (this is also the case with GM200 ofc, but 28nm was already really mature by the time Maxwell came along, which was what was so interesting about the huge die issues), and those yields improving (which we still don't really know who is doing what unfortunately. It seems like a dual source with Samsung and Gloflo, but we haven't heard anything like the A9 disparity between Samsung and TSMC, which might indicate it is only being sourced from identical processes even if at different plants).

 

Issues like overclocking headroom on later chips on the same process improving are only going to get more and more pronounced as the processes themselves get harder and harder to do.

 

Little Haswell Rant

Spoiler

 

I would argue that with the case of Haswell vs Haswell refresh the main difference isn't the process but the incredibly improved power-delivery of the chip itself (HARDWARE unlike AMD's software changes for the 390x). Well that and not using literally the worse TIM ever....

 

 

 

 

Anyways. I don't really care if they don't overclock that well... because value looks awesome, and I love the idea of a mainstream platform only from AMD (forcing Intel to address their market segmentation issues).

[cj09beira]

6 minutes ago, Curufinwe_wins said:

I actually agreed it is mostly the same case for everything.

 

Little Hawaii Rant

  Reveal hidden contents

But actually similar production time frame 290x and 390x, the 390x actually overclock marginally worse, but do do better with power consumption (marginally). This is 100% due to some changes in SOFTWARE power delivery, not anything else. But yea 390x to launch 290x is no contest.

 

That said, all Fiji gpus still seem to suck donkey dick at overclocking, and 14/16nm processes from non-intel seem to be hitting pretty rough class ceilings on frequency (even though 2100 MHz is no joke from Pascal, and 1400 or so is pretty darn good from Polaris, I have yet to see any significant improvements there.)

 

 

The real thing to realize though is the overclocking headroom to be gained as a process matures is 100% not about AMD, but about the companies manufacturing the chips (this is also the case with GM200 ofc, but 28nm was already really mature by the time Maxwell came along, which was what was so interesting about the huge die issues), and those yields improving (which we still don't really know who is doing what unfortunately. It seems like a dual source with Samsung and Gloflo, but we haven't heard anything like the A9 disparity between Samsung and TSMC, which might indicate it is only being sourced from identical processes even if at different plants).

 

Issues like overclocking headroom on later chips on the same process improving are only going to get more and more pronounced as the processes themselves get harder and harder to do.

 

Little Haswell Rant

  Hide contents

 

I would argue that with the case of Haswell vs Haswell refresh the main difference isn't the process but the incredibly improved power-delivery of the chip itself (HARDWARE unlike AMD's software changes for the 390x). Well that and not using literally the worse TIM ever....

 

 

 

 

 

Anyways. I don't really care if they don't overclock that well... because value looks awesome, and I love the idea of a mainstream platform only from AMD (forcing Intel to address their market segmentation issues).

the best thing would be for amd to gain a lot of market share and intel take time to respond, that way they have more money for R/D which right now they must be sufuring from lack of funds i bet.

its easy to forget this is a small company against two huge companies 

[Belgarathian]

10 hours ago, DocSwag said:

It's a little low but not horrible considering it's an 8 core. If I recall when Intel did an ln2 demo for the 4790k they only got to 5.5 ghz yet the 4790k still could reach 4.8 ghz or so on water.

My stable 5GHz OC on 4790K

 

[dexT]

18 minutes ago, Belgarathian said:

My stable 5GHz OC on 4790K

 

Lock the BCLK at 100.00 and it will be.

[done12many2]

27 minutes ago, Belgarathian said:

My stable 5GHz OC on 4790K

 

That's pretty stout.  What does it score in Cinebench R15 at that clock speed?

[PlayStation 2]

Not bad, not bad at all.

I just want to see how others manage with this, and (not exactly pertaining to this) how the R7 1700 performs.

[BuckGup]

17 hours ago, Simon771 said:

1,8V for 5,1GHz ... so we can forget about going to 5,0GHz on water cooling

Hopefully 4,5GHz will be doable somehow.

If AMD has taught us anything it's clock speeds don't matter. They sold garbage CPUs by showing off 8 cores and 5Ghz!!! Yet a quad core at 3.8Ghz on Intel's side beat it

[Belgarathian]

30 minutes ago, done12many2 said:

 

That's pretty stout.  What does it score in Cinebench R15 at that clock speed?

1015cb is my highest. Usually between 990 - 1010.

 

6 minutes ago, SamStrecker said:

If AMD has taught us anything it's clock speeds don't matter. They sold garbage CPUs by showing off 8 cores and 5Ghz!!! Yet a quad core at 3.8Ghz on Intel's side beat it

Clock speeds do matter, but if your IPC is terrible then your single thread perf. will still be terrible (clock speed * IPC). 

 

[DocSwag]

3 hours ago, Belgarathian said:

My stable 5GHz OC on 4790K

 

Yeah but Intel did have trouble getting to 5.5 on LN2. Probably after 5 ghz it just gets really difficult to go farther.

[TechGod]

9 hours ago, AnonymousGuy said:

So basically it's a dog.  6950X, 25% more cores, 5.325Ghz in Cinebench on LN2.

 

Paired with their shitty 100mhz turbo boost whatever they're calling it, Ryzen ain't going to be overclocking for shit.

Are you serious or high?

[vanished]

I guess I'm possibly not among company in this thread, but I couldn't care less what it will hit under LN2 or some other extreme test that people do just for the sake of it.  I want to know how it will run overclocked on air or a reasonable water setup - you know, something us normies might actually do  

[Drak3]

10 minutes ago, Ryan_Vickers said:

I guess I'm possibly not among company in this thread, but I couldn't care less what it will hit under LN2 or some other extreme test that people do just for the sake of it.  I want to know how it will run overclocked on air or a reasonable water setup - you know, something us normies might actually do  

No, I don't give a shit about a single LN2 run on boards that might not be stable platforms for Ryzen OCs yet. And I won't give a shit about LN2 until an enthusiast can use one for their main rig as effectively as a higher end custom water loop.

 

I'm also not looking at this LN2 run as the be all, end all of how well Ryzen will overclock.

[Bananasplit_00]

9 hours ago, vorticalbox said:

which kinda sucks seeing as I can get 5ghz at 1.5 on my 8350.

not really seeing as they will still smash that CPU hard even at stock because of the improved IPC

 

[TVwazhere]

(I apologize, I dont know if pcgamer qualifies as a creditable source)

 

http://www.pcgamer.com/amd-ryzen-7-1800x-hits-52-ghz-breaks-benchmarking-record/?utm_content=buffer05908&utm_medium=social&utm_source=facebook&utm_campaign=buffer_pcgamerfb

 

Reportedly, a multi-threaded run of benchmarking tool Cinebench R15 was completed on the Ryzen 7 1800X cooling at 5.2 GHz, achieving a score of 2449, beating the previous world record of 2410.  The screenshot of the test shows the CPU speed reaching 5201.07 MHz, thanks to a core voltage of 1.875V, the bus speed at 137.78 MHz and the multiplier up at 37.75x. Of course, this was done by professionals, and with the help of some liquid nitrogen cooling taking the temperature down to -200C. 

 

[NTDaws]

That's nice, but for all I know it could be fake.

[NTDaws]

I want to see benchmarks by eye.

[Guest Kloaked]

Just gonna point out a couple things.

 

"According to TweakTown, a YouTuber uploaded a video, which has since been taken down. Hexus has spotted a Vimeo link which remains active though, and you can see a brief screenshot of the benchmark results, as well as the moment the team breaks the record."

 

• TweakTown is full of shit mostly
• The video was deleted, which is suspicious
• TweakTown is full of shit