Post your Cinebench R20+15+R11.5+2003 Scores **Don't Read The OP PLZ**

[arnavvr]

1 minute ago, done12many2 said:

 

done12many2 -- 7900x -- 5 GHz -- Multi 2830cb / Single 225cb

 

 

I thought you got a 7920X?

[done12many2]

Just now, arnavvr said:

I thought you got a 7920X?

 

I did, but I had some concerns with memory performance on the bigger chip, so I went back to the 7900x. 

[rip]

what multi and single threaded scores should i expect from a sandy bridge xeon 1240? no oc and stuff, with a gtx 670 if that matters

[purple_rider]

17 hours ago, rip said:

what multi and single threaded scores should i expect from a sandy bridge xeon 1240? no oc and stuff, with a gtx 670 if that matters

You should get 500-700cb judging by other entries in the spreadsheet.

[SandMartian]

I built this PC with help from many videos from Linus and jumped in with both feet.

 

I am a visual effects artist and mainly run Cinema4D, and adobe stuff.    I am completely new to AMD.

 

Overall I am very happy with the build but have a feeling things could be running better.  My memory is supposed to

be 3600 but it look like it's only doing a little over 2000.  Various other bugs plague me and I hope to be able to navigate 

this forum and eventually get some answers.  

 

Thank you in advance for any observations and I am happy to share what I know as well.

 

 

That temp seems so low.  could this be right?  Is there an offset or something?  

So many questions about this build.

 

[rip]

13 hours ago, purple_rider said:

You should get 500-700cb judging by other entries in the spreadsheet.

multi or single threaded?

[Cheddle]

On 07/10/2017 at 4:53 PM, done12many2 said:

 

done12many2 -- 7900x -- 5 GHz -- Multi 2830cb / Single 225cb

 

 

Great chip!

 

mine certainly cant pass at those voltages... What vrin are you running?

[PianoPlayer88Key]

 

Sorry I didn't get the CPU-Z window (re-read the OP in this thread & found out I would have needed it - AFTER I had shut down the system running the benchmark).  Also I don't feel like re-running it right now, as you can guess from the not-so-1337 score I got.

 

Spoiler

Yeah, things might not be quite as they seem.  I was experimenting with some things.

 

CPU execution cap was set at 1%.  If I'm doing the math right, 1% of 0.76 GHz (760 MHz) is 7.6 MHz.

 

I had another VM running where I had Windows 10 running with 1 thread, an execution cap of 5%, and 256 MB RAM.  Windows actually would load, but I'm not sure if Cinebench would have loaded.  Although.... I wonder how low you COULD go on allocated resources and have it still load.  (Not that you could actually DO anything productive with it, of course...)

 

Somehow the original images got deleted.

 

P.S. I thought there had been some place where a few people had posted very low cinebench scores (like in the low double digits on R15), but I was unable to find it. (Maybe it was earlier in this thread, but idk.)  Also you probably don't need to bother putting this one in the spreadsheet.  You wouldn't expect a 4790K at 4 GHz to get a cinebench score of 4, would you?  

 

[zxsq]

New ryzen 7 1700x oc: 1711.

[purple_rider]

9 hours ago, rip said:

multi or single threaded?

Multi

[done12many2]

22 hours ago, Cheddle said:

Great chip!

 

mine certainly cant pass at those voltages... What vrin are you running?

 

I thought so. My first 7900x was damn good, but this one is even better. 

 

Keep in mind that it's now delidded so voltages are slightly better as a result of better cooling. 

 

By default, I think the Apex runs these chips at 1.95 input voltage. 

[Cheddle]

Just now, done12many2 said:

 

I thought so. My first 7900x was damn good, but this one is even better. 

 

Keep in mind that it's now delidded so voltages are slightly better as a result of better cooling. 

 

By default, I think the Apex runs these chips at 1.95 input voltage. 

Im on 1.9v Vrin IIRC but I was running it stock with no real difference - mine is delidded too - I can cinebench as low as 4.8ghz @ 1.23v but to get realbench stability I have to run 1.265v - so thats my daily setting.

 

I tried 5ghz cinebench at 1.28v and couldnt find stability. I havn't gone above 1.28v on this unit (1.32v was as high as I went on my first cpu to get 5ghz - and it died in two days so im real scared haha)

 

My CPU is the SL 4.6ghz bin 

[done12many2]

6 minutes ago, Cheddle said:

Im on 1.9v Vrin IIRC but I was running it stock with no real difference - mine is delidded too - I can cinebench as low as 4.8ghz @ 1.23v but to get realbench stability I have to run 1.265v - so thats my daily setting.

 

I'm still trying to learn the chip well enough, but I did a one hour pass on RealBench at 4.8 GHz with no offset at roughly 1.17 to 1.18v on the cores.  Here's a screen shot half way through. I'll get around to testing 4.9 at some point I'm sure. 

 

 

6 minutes ago, Cheddle said:

I tried 5ghz cinebench at 1.28v and couldnt find stability. I havn't gone above 1.28v on this unit (1.32v was as high as I went on my first cpu to get 5ghz - and it died in two days so im real scared haha)

 

My CPU is the SL 4.6ghz bin 

 

Considering that we're both running the chips delidded, it could just be a case of binning or cooling. 

 

What board are you running?

[Cheddle]

4 minutes ago, done12many2 said:

 

I'm still trying to learn the chip well enough, but I did a one hour pass on RealBench at 4.8 GHz with no offset at roughly 1.17 to 1.18v on the cores.  Here's a screen shot half way through. I'll get around to testing 4.9 at some point I'm sure. 

 

 

 

Considering that we're both running the chips delidded, it could just be a case of binning or cooling. 

 

What board are you running?

X299 Taichi. 

 

binning will definitly be the difference here.

 

Im also running 3000mhz ring clock with stock voltage

 

What vccsa/vccio/vtt/dram voltage are you using to get those memory clocks? b-die I assume? whats the rated speed/voltage on the kit?

 

personally im at stock vccsa/vccio (force stock, not the XMP defaults), 690mv vtt and 1.38v dram. kit is b-die rated at 3600-16-16-16 and im running it at 4000-cl17-18-18-40-cr1-350 from memory.

[done12many2]

10 hours ago, Cheddle said:

X299 Taichi. 

 

binning will definitly be the difference here.

 

Im also running 3000mhz ring clock with stock voltage

 

Despite what you see in the previous RealBench screen shot, I am now running 3000 MHz (or higher depending on what I'm doing) cache.  That screen shot was just some initial testing after a delid with nothing more than CPU OC and default XMP loaded (I think ).

 

I did check to see what the Apex input voltage defaults to and it's 1.89v.

 

Quote

What vccsa/vccio/vtt/dram voltage are you using to get those memory clocks? b-die I assume? whats the rated speed/voltage on the kit?

 

I am using a G.Skill TZ 4 x 8 3200 MHz kit (14-14-14-34-2T XMP).

 

When I run XMP speeds, I just manually set VCCSA/VCCIO/VTT all to .9v and use the XMP default 1.35v.

 

When I run 4000 MHz 16-17-17-39-1T with tight secondaries, I run BIOS values of .925v VCCSA / .9v VCCIO / .9v VTT / 1.39v DRAM.  This results in a perfectly stable mem overclock with high bandwidth and low latency.  I also run 3200 MHz cache most of the time when I run my memory at 4000 MHz.

 

Here's a screen shot that has my timings (primaries and secondaries) along with performance numbers if that's any help.

 

[Ground]

Something seems to be wrong with either my L5640 or my board. It doesn't allow me to raise the BCLK manually, I can only raise it using the hardware switches on the board. Well, 3.4 GHz out of a 2.26 GHz chip isn't terrible, but still kinda disappointing...

I'll probably give my e5620 another run now, as I'm already switch CPU's back and forth anyways...

[dexT]

1 hour ago, Ground said:

Something seems to be wrong with either my L5640 or my board. It doesn't allow me to raise the BCLK manually, I can only raise it using the hardware switches on the board. Well, 3.4 GHz out of a 2.26 GHz chip isn't terrible, but still kinda disappointing...

That should be 4.2, 21x200.

 

You should make it run the OC freq fulltime when benchmarking. The time it takes to turbo up hurts the score.

[Ground]

3 minutes ago, dexT said:

That should be 4.2, 21x200.

 

You should make it run the OC freq fulltime when benchmarking. The time it takes to turbo up hurts the score.

 

My board doesn't support running all cores at the max multiplier. The max I can use for benching multicore is 19...

[dexT]

22 minutes ago, Ground said:

My board doesn't support running all cores at the max multiplier. The max I can use for benching multicore is 19...

That sucks. Even in full manual mode?

[Ground]

5 minutes ago, dexT said:

That sucks. Even in full manual mode?

There is only one mode, it's basically the cheapest board you can overclock with for x58. But apparently I managed to get overclockig working again properly (at least my e5620 booted at 4 GHz, so its time to swap the L5640 in and overclock it properly this time

Seems like it was a faulty BIOS battery. The 32nm Quad Core that stayed cool at 1.5V under air - now only if it would overclock properly...

[dexT]

6 minutes ago, Ground said:

There is only one mode, it's basically the cheapest board you can overclock with for x58. But apparently I managed to get overclockig working again properly (at least my e5620 booted at 4 GHz, so its time to swap the L5640 in and overclock it properly this time

Seems like it was a faulty BIOS battery. The 32nm Quad Core that stayed cool at 1.5V under air - now only if it would overclock properly...

Can you disable C1E and TM to lock the max freq?

[Ground]

8 minutes ago, dexT said:

Can you disable C1E and TM to lock the max freq?

Well, let's see. I have the E5620 dialed in at 240x19, lets see if it actually runs... The highest I had before was 230x18, so no matter what, it'll be an improvement.

[dexT]

6 minutes ago, Ground said:

Well, let's see. I have the E5620 dialed in at 240x19, lets see if it actually runs... The highest I had before was 230x18, so no matter what, it'll be an improvement.

Open task manager processes and kill your realtime virus scanner if you have one. Right click Cinebench in processes and set priority to realtime or high if you can't do RT. Run it four times without closing the R15 window letting it cool fully between runs and use the best one. Screenshot each run and close cpuz before each run.

[Cheddle]

10 hours ago, done12many2 said:

 

Despite what you see in the previous RealBench screen shot, I am now running 3000 MHz (or higher depending on what I'm doing) cache.  That screen shot was just some initial testing after a delid with nothing more than CPU OC and default XMP loaded (I think ).

 

I did check to see what the Apex input voltage defaults to and it's 1.89v.

 

 

I am using a G.Skill TZ 4 x 8 3200 MHz kit (14-14-14-34-2T XMP).

 

When I run XMP speeds, I just manually set VCCSA/VCCIO/VTT all to .9v and use the XMP default 1.35v.

 

When I run 4000 MHz 16-17-17-39-1T with tight secondaries, I run BIOS values of .925v VCCSA / .9v VCCIO / .9v VTT / 1.39v DRAM.  This results in a perfectly stable mem overclock with high bandwidth and low latency.  I also run 3200 MHz cache most of the time when I run my memory at 4000 MHz.

 

Here's a screen shot that has my timings (primaries and secondaries) along with performance numbers if that's any help.

 

thanks! thoise will be VERY useful 

 

0.9v VTT sounds quite high? how long have you ran that without issues? mines about 0.65v stock from memory

 

 

[done12many2]

40 minutes ago, Cheddle said:

thanks! thoise will be VERY useful 

 

0.9v VTT sounds quite high? how long have you ran that without issues? mines about 0.65v stock from memory

 

 

 

Truthfully, I haven't exactly determined highs for x299 yet. What do you run on VTT? I'll give it a shot. Thanks