Intel Core i7-10700

[ppn]

There was broadwell 14nm, almost 6 years now, IPC did not improve significantly. just cosmetic changes like DDR4.

tsmc 7nm offers only 2x density. so intels 1,5x is not bad at all,
judging by how many times they improved the power, but that last thing doesn't seem to hold true.

"Compared to all other "14 nm nodes", Intel's process is the densest and considerably so, with >1.5x raw logic density. "
"They improved on their original process with a second process, "14nm+", offering 12% higher drive current at lower power "
"A third improved process, "14nm++", is set to begin in late 2017 and will further allow for +23-24% higher drive current for 52% less power vs the original 14nm process. The 14nm++ process also appear to have slightly relaxed poly pitch of 84 nm (from 70 nm) "

 

[danbert2000]

I find the power consumption pages misleading, because for all the Ryzen 3xxx processors, an x570 board is used, which drives up power consumption by 30 W or more. Seeing as these processors run just as fast on B450 or X470 boards, it would make sense to see what the power consumption is with a chipset that doesn't take so much power. All it does is make it look like Intel is closer in power efficiency than they actually are. At least some testing with the older boards as a comparison would be good. The X570 chipset is going to be replaced by the B550 and X590, but these power consumption figures will essentially be the "historical record" for this website, making very efficient AMD processors look less so based solely on the configuration of the test system.

 

[W1zzard]

I find the power consumption pages misleading, because for all the Ryzen 3xxx processors, an x570 board is used, which drives up power consumption by 30 W or more. Seeing as these processors run just as fast on B450 or X470 boards, it would make sense to see what the power consumption is with a chipset that doesn't take so much power. All it does is make it look like Intel is closer in power efficiency than they actually are. At least some testing with the older boards as a comparison would be good. The X570 chipset is going to be replaced by the B550 and X590, but these power consumption figures will essentially be the "historical record" for this website, making very efficient AMD processors look less so based solely on the configuration of the test system.

Same argument for Intel chipsets I guess? This Z490 does use a lot of power too

 

[danbert2000]

A comparison of load minus idle would potentially give more information about a processor's power consumption than a raw reading with no "control." Since you are trying to compare processor power consumption, I would expect some desire to tease out the actual processor's contribution. It is a good point that the Z490 uses power too, but at what point are we comparing motherboard chipsets instead or in addition to processors?

The full system draw is useful for people trying to size their PSUs, but not useful for comparing single threaded and multi threaded power efficiency. Especially since AMD users can use a B board without sacrificing performance.

Even a note that X570 power consumption is a contributor to the figures would be appreciated.

 

[RandallFlagg]

You can't just choose based on chipset, chipset <> power draw. I think it's totally valid to use the fastest components available for a given platform (within spec, not overclocked) to test a new CPU with. That would include power draw, as the fastest motherboards often take the most power.

Case in point, there's a 50W difference between these Z490 boards under load. And guess what? The fastest board, also draws the most power. These are with a 10900K :

 

[Elysium]

Presumably the i9-10900 could be similarly power-unlocked to provide a tangible gain in perf to catch up to the K variant whilst being a few dozen dollars cheaper? That might just about rejuvenate my interest in Comet Lake...

 

[yeeeeman]

Yet you don't seem to have any bias, towards Intel, is that what you're claiming? I highlighted the specific points of your post ~ which you should know are incorrect but whatever floats your boat

And before you go on to nitpick on this reply, why don't you post evidence to backup your initial claim that everything's (nearly) the same as it was 6 years back? For instance there's no SKL in 2014, so no backtracking room there.

Taken from wikipedia: "Broadwell: 14 nm shrink of the Haswell microarchitecture, released in September 2014". The process is 6 years old, not SKL.
Lets close this here because your clearly not being able to read and just try to invalidate my opinions.
It is fine, you'll have yours and I'll have mine. But my opinions are backed up by facts.

Presumably the i9-10900 could be similarly power-unlocked to provide a tangible gain in perf to catch up to the K variant whilst being a few dozen dollars cheaper? That might just about rejuvenate my interest in Comet Lake...

Yes, that is true.

A comparison of load minus idle would potentially give more information about a processor's power consumption than a raw reading with no "control." Since you are trying to compare processor power consumption, I would expect some desire to tease out the actual processor's contribution. It is a good point that the Z490 uses power too, but at what point are we comparing motherboard chipsets instead or in addition to processors?

The full system draw is useful for people trying to size their PSUs, but not useful for comparing single threaded and multi threaded power efficiency. Especially since AMD users can use a B board without sacrificing performance.

Even a note that X570 power consumption is a contributor to the figures would be appreciated.

I think it is safe to say that the 10700 is within the same range of power consumption as the 3700x, even including the chipset mismatch. Be assured that Z490 is also a power hog since it has a lot of stuff integrated inside, from TB3 to WF6... So again, the same conclusion holds, the 10th gen is a very nice surprise in terms of efficiency, especially compared to 7nm parts from AMD.

 

[R0H1T]

Lets close this here because your clearly not being able to read and just try to invalidate my opinions.

Invalidate what exactly, your opinion or made up facts? Sure because you made a false comparison. From your own quote ~

Taken from wikipedia: "Broadwell: 14 nm shrink of the Haswell microarchitecture, released in September 2014". The process is 6 years old, not SKL.

6 years from your comment would imply June 2014, not September. There have been at least 2 major revisions to 14nm & yet you made it sound as if that's what AMD is competing against, here you go try to spin this any way you want to ~

A 14nm process, 6 years old now gives you the same performance

 

[mahanddeem]

I have a question please about the 10700 non-k.
If you use it in a Z490 MB:
1. Can you activate XMP (my ram is 3200 16-8-18-36)
2. Can you activate sync all cores at turbo clock? Or in Asus words the MCE (core enhancement) syncing all cores to like 4.6GHz?

Thanks

 

[W1zzard]

I have a question please about the 10700 non-k.
If you use it in a Z490 MB:
1. Can you activate XMP (my ram is 3200 16-8-18-36)
2. Can you activate sync all cores at turbo clock? Or in Asus words the MCE (core enhancement) syncing all cores to like 4.6GHz?

1) yes, works exactly the same as on a K CPU
2) no

 

[mahanddeem]

1) yes, works exactly the same as on a K CPU
2) no

Thanks

 

[Melvis]

@W1zzard Clock speeds are wrong in your CPU chart first page for the Ryzen 3600 again.

 

[mahanddeem]

What would you get between these 2? 10600k or 10700 for gaming at 1440p with a 1080ti.
Thinking to upgrade my 7700k

 

[W1zzard]

@W1zzard Clock speeds are wrong in your CPU chart first page for the Ryzen 3600 again.

Fixed in all reviews and in my template, thanks

 

[Dirtdog]

You can't just choose based on chipset, chipset <> power draw. I think it's totally valid to use the fastest components available for a given platform (within spec, not overclocked) to test a new CPU with. That would include power draw, as the fastest motherboards often take the most power.

Case in point, there's a 50W difference between these Z490 boards under load. And guess what? The fastest board, also draws the most power. These are with a 10900K :

View attachment 157734

May I ask where that graph is from? The ASRock boards look to have pretty high idle power draw. I've been hoping to see a review of the Extreme4 to see if it still has the same excessive power draw issue its Z390 predecessor had.

 

[RandallFlagg]

May I ask where that graph is from? The ASRock boards look to have pretty high idle power draw. I've been hoping to see a review of the Extreme4 to see if it still has the same excessive power draw issue its Z390 predecessor had.

It's from Vortez : https://www.vortez.net/

All of their new MB reviews do overall power consumption graphs. Also the IO performance graphs are revealing. There are some significant differences in performance between motherboards that affect overall system performance, but rarely are CPU focused activities impacted much (games, synthetics like 'cinebench', and so on) so they redid their benchmarks to show the areas where there are real differences. Power and IO are big differentiators.

 

[Dirtdog]

It's from Vortez : https://www.vortez.net/

All of their new MB reviews do overall power consumption graphs. Also the IO performance graphs are revealing. There are some significant differences in performance between motherboards that affect overall system performance, but rarely are CPU focused activities impacted much (games, synthetics like 'cinebench', and so on) so they redid their benchmarks to show the areas where there are real differences. Power and IO are big differentiators.

Thanks, looks like it's this page: https://www.vortez.net/articles_pages/asrock_z490_taichi_review,5.html

So they say the Z490 Taichi has an idle power draw of 87w which is huge, but the same board reviewed by another site puts it at 70w, that's with a 10900K: https://www.guru3d.com/articles_pages/asrock_z490_taichi_review,7.html

Vortez has the Asus ROG Max II Extreme idle power as 61w (substantially lower than they cite for the Z490 Taichi) but the Guru3D article above has that board using 75w idle.

Tweaktown has the two respective boards using 70w and 76w: https://www.tweaktown.com/reviews/9473/asrock-z490-taichi-intel-atx-motherboard/index12.html

So Vortez having them using 87w vs 61w lacks credibility and draws into question their Extreme4 figures (and that board doesn't seem to be reviewed on their site in its own right either).

I notice in their conclusion and summary that they don't seem to mention the high idle power draw, which is 20-30w higher than every other board in the same class and would represent a noteworthy flaw, but they don't seem fazed by it which is very strange.

 

[W1zzard]

which is very strange.

Those storage results cant be right, especially m.2

 

[RandallFlagg]

Thanks, looks like it's this page: https://www.vortez.net/articles_pages/asrock_z490_taichi_review,5.html

So they say the Z490 Taichi has an idle power draw of 87w which is huge, but the same board reviewed by another site puts it at 70w, that's with a 10900K: https://www.guru3d.com/articles_pages/asrock_z490_taichi_review,7.html

Vortez has the Asus ROG Max II Extreme idle power as 61w (substantially lower than they cite for the Z490 Taichi) but the Guru3D article above has that board using 75w idle.

Tweaktown has the two respective boards using 70w and 76w: https://www.tweaktown.com/reviews/9473/asrock-z490-taichi-intel-atx-motherboard/index12.html

So Vortez having them using 87w vs 61w lacks credibility and draws into question their Extreme4 figures (and that board doesn't seem to be reviewed on their site in its own right either).

I notice in their conclusion and summary that they don't seem to mention the high idle power draw, which is 20-30w higher than every other board in the same class and would represent a noteworthy flaw, but they don't seem fazed by it which is very strange.

The exact numbers are different between sites, but this can depend on the exact GPU, SSD, number of storage drives, and so on which are not the same for every site.

That said, the pattern of difference for example between Vortez and Tweaktown is similar and shows the same general conclusions.

Below are charts from tweaktown showing a full system power of 16W between ASRock Z490 Taichi (275W) and Gigabyte Z490 Aorus Xtreme (291W), and a 67W difference between the Aorus Xtreme and the Supermicro Z490.

Tweaktown also shows a similar pattern regarding disk IO performance (2nd graph).

I'm not real sure how you could look at these other sites, see charts like this, and not see that there is a big difference in both power draw and IO performance on difference motherboards. The numbers don't need to be exactly the same to see a pattern.

I suspect the difference has to do with VRM design for power, and probably BIOS for IO.

Disk IO :

USB performance:

 

[md2003]

@W1zzard thanks for the review. I do have one specific question, regarding the set muliplier in overclocking scenarios, specifically when blck is set higher than 100. Is it an asus specific feature to actually set the x47 multiplier in such scenario? In msi z490 tomahawk only x46 multiplier is set/activated, no matter the settings, thus a max freq of 4.7+Ghz is achieved. With less cores enabled, i.e. 2 cores or 4 cores 4.9Ghz+ and 4.8Ghz+ are achieved, but with all cores enabled the x46 multiplier prevails.
As example, a cpu validation, where max multi is shown as 47 but the x46 is used only: https://valid.x86.fr/5su2jm

 

[W1zzard]

only x46 multiplier is set/activated, no matter the settings

At all times? Are you sure? Try with light loads, not stress test. Also running into power limit will lower your clocks. And some monitoring software itself creates additional load on the system, enough for the CPU to not boost to highest Clock

 

[md2003]

At all times? Are you sure?

When bclk is higher than 100, the multiplier is locked to x46, no matter the system load. With everything at stock, without xmp, it uses sporadically x47 and x48 too on all cores but not on high loads (so every core boosts up to 4.8GHz but not all together, which is as expected). Last but not least, in bios you can set all cores or turbo ratio per core up to x48 (it is even accepted in bios all cores simultaneously to use x48) but in windows sticks to x46.

 

[W1zzard]

Interesting, i wasn’t aware of any correlation between bclk and multi.. on holidays for another week, will check on it when back. If you don’t hear from me, Ping me, so i won’t forget

 

[md2003]

Interesting, i wasn’t aware of any correlation between bclk and multi.. on holidays for another week, will check on it when back. If you don’t hear from me, Ping me, so i won’t forget

Well, i think XMP is to blame for. I set ram timings, volts etc manually, and now i even get this x48 on all cores using higher than 100 bclk (i attach ss from bios configuration). Single core performance is where it should have been now i guess, https://valid.x86.fr/1d5i0d. Of course, i would be happy to read more from your findings.

 

[W1zzard]

Maybe this is some kind of bios bug you’reseeing?