Discussion Alder Lake - Builders Thread

Page 11 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

AdamK47

Lifer
Oct 9, 1999
15,223
2,844
126
This thread is for those of us that own or are looking to get an Alder Lake CPU.

Bought mine on release day. 12900K. Memory selection was limited to only Crucial DDR5 4800 at the Micro Center I went to. Bought two 2 x 8GB kits. Motherboard is a Gigabyte Z690 Aorus Master. I did buy a 360mm Lian Li Galahad 360mm AIO. Ended up returning it since the Lian Li Socket 1700 backplate that I bought separately worked well with my NZXT Kraken X73.

46RRrSN.jpg
 
Last edited:

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,514
136
OK, going to try and minimize this output, but make it available. So the first 16 lines is the 7B12 64 core EPYC units for today and the second is a 3950x. The 3950x does .59 points per minute, and the 7B12 does .53. But the 7B4 does 4 times as many a minute. This is Rosetta. And this result I would consider to be very comparable. Not primegrid uses a lot more floating point, so its results could be different. The 5950x just screams at primegrid.

Does this give you an idea what I will be doing ? Oh, and adding points per watt. And it will be running linux, so most of the benchmarks and tests you guys use will not work.

4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:43:59 UTC10,586.1410,553.4393.99
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:46:31 UTC10,777.1510,729.6093.63
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:46:31 UTC10,637.6210,585.4598.64
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:04 UTC10,686.3610,639.6496.69
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:48:11 UTC10,903.3810,850.80103.52
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:43:59 UTC10,616.4410,567.1479.66
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:04 UTC10,773.2210,713.60104
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:38 UTC10,864.0410,817.2097.64
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:04 UTC10,822.9810,785.62103.01
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:04 UTC10,723.8010,673.1396.95
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:46:31 UTC10,707.5010,648.1493.47
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:43:59 UTC10,828.1710,773.4695.76
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:43:22 UTC10,809.6210,769.9897.2
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:42:37 UTC10,786.4410,745.2291.15
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:42:37 UTC10,813.9410,757.2592.48
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:43:59 UTC10,838.2810,788.70101.7
10760.94​
10712.4​
96.21813​
0.538916​
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 16:17:50 UTC28,377.2527,522.88217.86
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC29,027.9328,159.06313.16
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC29,070.1528,202.94321.03
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 19:44:03 UTC28,769.9427,859.06268.17
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC29,387.3528,457.04304.73
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC29,225.3428,337.09256.06
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC28,123.7527,314.34302.71
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 20:45:39 UTC29,289.6228,366.59293.23
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 16:17:50 UTC29,824.7428,938.16231.95
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC29,036.8228,038.41246.8
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC28,550.4927,719.59200.01
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 16:17:50 UTC28,650.2927,749.10372.36
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC28,705.8727,821.00319.7
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC28,328.9027,433.25278.85
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 20:45:39 UTC28,874.2427,966.73265.16
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC28,786.2427,925.64215.07
28876.81​
27988.18​
275.4281​
0.590452​
 
Last edited:

Hitman928

Diamond Member
Apr 15, 2012
5,282
7,915
136
OK, going to try and minimize this output, but make it available. So the first 16 lines is the 7B12 64 core EPYC units for today and the second is a 3950x. The 3950x does .59 points per minute, and the 7B12 does .53. But the 7B4 does 4 times as many a minute. This is Rosetta. And this result I would consider to be very comparable. Not primegrid uses a lot more floating point, so its results could be different. The 5950x just screams at primegrid.

Does this give you an idea what I will be doing ? Oh, and adding points per watt. And it will be running linux, so most of the benchmarks and tests you guys use will not work.

4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:43:59 UTC10,586.1410,553.4393.99
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:46:31 UTC10,777.1510,729.6093.63
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:46:31 UTC10,637.6210,585.4598.64
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:04 UTC10,686.3610,639.6496.69
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:48:11 UTC10,903.3810,850.80103.52
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:43:59 UTC10,616.4410,567.1479.66
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:04 UTC10,773.2210,713.60104
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:38 UTC10,864.0410,817.2097.64
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:04 UTC10,822.9810,785.62103.01
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:47:04 UTC10,723.8010,673.1396.95
4 Mar 2022, 6:48:40 UTC4 Mar 2022, 21:46:31 UTC10,707.5010,648.1493.47
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:43:59 UTC10,828.1710,773.4695.76
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:43:22 UTC10,809.6210,769.9897.2
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:42:37 UTC10,786.4410,745.2291.15
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:42:37 UTC10,813.9410,757.2592.48
4 Mar 2022, 6:48:39 UTC4 Mar 2022, 21:43:59 UTC10,838.2810,788.70101.7
10760.94​
10712.4​
96.21813​
0.538916​
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 16:17:50 UTC28,377.2527,522.88217.86
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC29,027.9328,159.06313.16
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC29,070.1528,202.94321.03
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 19:44:03 UTC28,769.9427,859.06268.17
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC29,387.3528,457.04304.73
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC29,225.3428,337.09256.06
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC28,123.7527,314.34302.71
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 20:45:39 UTC29,289.6228,366.59293.23
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 16:17:50 UTC29,824.7428,938.16231.95
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC29,036.8228,038.41246.8
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC28,550.4927,719.59200.01
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 16:17:50 UTC28,650.2927,749.10372.36
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC28,705.8727,821.00319.7
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 18:41:10 UTC28,328.9027,433.25278.85
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 20:45:39 UTC28,874.2427,966.73265.16
3 Mar 2022, 23:24:08 UTC4 Mar 2022, 17:50:44 UTC28,786.2427,925.64215.07
28876.81​
27988.18​
275.4281​
0.590452​

You could run Blender in Linux. It's not as easy to run as Cinebench (one click and go) but it's not hard to run and you could use it as a sanity check against reviewer results. There's also Phoronix Test Suite. It's not quite as straight forward but not overly difficult.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,514
136
You could run Blender in Linux. It's not as easy to run as Cinebench (one click and go) but it's not hard to run and you could use it as a sanity check against reviewer results. There's also Phoronix Test Suite. It's not quite as straight forward but not overly difficult.
But all those tests are available, and everybody knows them, I do not contest most of what I have seen benchmarked, thats why I am trying this. That, and I could not resist $313 for something very close to the $400 5900x (close in many multu-threaded benchmarks to the 12700k) and the regular $400 for the 12700k.
 
  • Like
Reactions: Drazick
Jul 27, 2020
16,312
10,335
106
OK, going to try and minimize this output, but make it available. So the first 16 lines is the 7B12 64 core EPYC units for today and the second is a 3950x. The 3950x does .59 points per minute, and the 7B12 does .53. But the 7B4 does 4 times as many a minute.
This seems counter-intuitive. 7B12 with 4 times as many cores as 3950X did less work. 7B4 is a typo? Googling Epyc 7B4 doesn't seem to indicate this model exists.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,514
136
This seems counter-intuitive. 7B12 with 4 times as many cores as 3950X did less work. 7B4 is a typo? Googling Epyc 7B4 doesn't seem to indicate this model exists.
Rosetta is odd. But as I said, it gets almost the same points/minute for x number of jobs. The 3950x does 1/4 the number of jobs. But in this case the run times are like 3 times linger for the 3950x, but 3 times the points.The run times are because of the way Rosetta works.

And its 7B12, not 7B4. See here:


But it only does 2.6 ghz fully loaded. The 3.3 is max boost.
 
Last edited:
  • Like
Reactions: Drazick
Jul 27, 2020
16,312
10,335
106
I understand the runtimes being longer for 3950X as it has fewer cores so they have to work longer but 3 times the number of points? Is this some Rosetta scheme to get people with fewer cores to do jobs so they feel good that they are getting more points?
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,514
136
I understand the runtimes being longer for 3950X as it has fewer cores so they have to work longer but 3 times the number of points? Is this some Rosetta scheme to get people with fewer cores to do jobs so they feel good that they are getting more points?
Rosetta is weird... The run times are all over the place. The 7B12 is running 128 tasks at a time, but the 3950x is doing 32 at a time. The run times are set by Rosetta, but as you can see, adjusted points based on that. I want to use WCG as my DC benchmark for my 12700k, its not odd, but won't be back up until 22 April. I could try primegrid.
 
  • Like
Reactions: Drazick

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,514
136
Any data points for that, say comparing 3950X to 5950X?
Running now. But the results won't work. For some reason, the 3950x has only 1 tasks, so its running faster and getting it done quicker. The 5950x is only running 11, but they are still running slower. I need good old WCG to really tell.

Edit: @igor_kavinski . When I get the 12700k setup, I will start a thread in the DC forum for the benchmark comparison. I don't want to muck up the ADL builders thread, except details on my 12700k build, OK ?
 
Last edited:
  • Like
Reactions: Drazick

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,514
136
OK, I just added G.Skill RipJaws V Series 32GB (2 x 16GB) 288-Pin SDRAM DDR4 4000 (PC4-32000) CL18-22-22-42 1.40V Dual Channel Desktop Memory Model F4-4000C18D-32GVK and the recommended ID-COOLING SE-226-XT Black CPU Cooler LGA1700 CPU Air Cooler 6 Heatpipes All-Black CPU Fan 120mm PWM Fan
and EVGA 850 GQ, 80+ GOLD 850W.

So the build is complete except the case, which I have and its crap, but it works.
 

Justinbaileyman

Golden Member
Aug 17, 2013
1,980
249
106
So I got what was supposed to be my Z690 Aorus Master on Sat. from Bestbuy but delivered from UPS. First off it did not ship in a protective shipping box but rather a Shrink wrapped brown bag. Second this is not a Motherboard but instead they sent me a damn $3000.00 Razor laptop..

I contacted them but they wanted me to pay almost $75 out of my own pocket to ship it back and then wait up to a month or more for processing and return of my payment. I ended up driving like an hour away on a Sunday just to return this crap. and get my money back..

So now I am with out any motherboard again for my new build. Now my plan was to slowly piece together a 12900k build or go ultra cheap on the CPU end with say a 12400 and upgrade to the top end Raptorlake when released. So do you guys think I should try again and repurchase the motherboard from Bestbuy or should just upgrade my current 3950xrig to a 5950x?? What would you guys do??
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,560
14,514
136
So I got what was supposed to be my Z690 Aorus Master on Sat. from Bestbuy but delivered from UPS. First off it did not ship in a protective shipping box but rather a Shrink wrapped brown bag. Second this is not a Motherboard but instead they sent me a damn $3000.00 Razor laptop..

I contacted them but they wanted me to pay almost $75 out of my own pocket to ship it back and then wait up to a month or more for processing and return of my payment. I ended up driving like an hour away on a Sunday just to return this crap. and get my money back..

So now I am with out any motherboard again for my new build. Now my plan was to slowly piece together a 12900k build or go ultra cheap on the CPU end with say a 12400 and upgrade to the top end Raptorlake when released. So do you guys think I should try again and repurchase the motherboard from Bestbuy or should just upgrade my current 3950xrig to a 5950x?? What would you guys do??
Just so I don't have to go way back and find your old posts. What do you do with the machine again ? The 12900k and the 12700k are not that different in many areas, just 4 E cores. My 12700k is due anytime today, and I might be able to do a few comparisions. For Gaming, the 12700k is almost identical to the 12900k, except uses way less power. For other things, the 5950 wins most things. But the 3950x is so close to the 5950x in a log of things, its not worth upgrading.

So again, what are you going to be doing with this computer ?
 
  • Like
Reactions: Drazick

Justinbaileyman

Golden Member
Aug 17, 2013
1,980
249
106
Just so I don't have to go way back and find your old posts. What do you do with the machine again ?
I do some gaming , but mostly do lots and lots of video and audio editing / encoding and compression. I also am a heavy browser tab user and have more then several dozen tabs open at a time. Most the time I have no less then a dozen or so apps or programs running in the back ground. Also use to watch and stream media.. That is the basics LOL
 

dullard

Elite Member
May 21, 2001
25,066
3,415
126
I do some gaming , but mostly do lots and lots of video and audio editing / encoding and compression. I also am a heavy browser tab user and have more then several dozen tabs open at a time. Most the time I have no less then a dozen or so apps or programs running in the back ground. Also use to watch and stream media.. That is the basics LOL
One of the big problems with Alder Lake is background tasks. The way Windows assigns tasks is that the background tasks go to the efficient cores and the foreground tasks go to the performance cores. That normally works well, most people want the foreground software to be fast and responsive. But, that means if you get one of the Alder Lakes with only 4 E cores, then all of those dozen or so apps you have in your background have to share the 4 weakest cores on your CPU. They could come crawling nearly to a halt. Thus, stay far away from the 12600K, 12700, or 12700K for that type of use (and any variant of those 3 chips).

For that reason, I'd personally go with the 12500 if you can then jump to Raptor Lake if the 12500 is not powerful enough. The 12500 is to me the sweet spot in Intel CPU value. Then once Raptor Lake comes around with enough E cores, that will solve the background task problem.
 
  • Like
Reactions: igor_kavinski
Jul 27, 2020
16,312
10,335
106
I do some gaming , but mostly do lots and lots of video and audio editing / encoding and compression. I also am a heavy browser tab user and have more then several dozen tabs open at a time. Most the time I have no less then a dozen or so apps or programs running in the back ground. Also use to watch and stream media.. That is the basics LOL
TBH, if you upgraded the RAM on your current system to 128GB, you might find it completely acceptable for your usage scenarios. Unless you really want shorter waiting times when video encoding/editing, I think spending more on RAM would help you more.
 
Jul 27, 2020
16,312
10,335
106
Thus, stay far away from the 12600K, 12700, or 12700K for that type of use (and any variant of those 3 chips).

For that reason, I'd personally go with the 12500 if you can then jump to Raptor Lake if the 12500 is not powerful enough.
Or disable E-cores if you go with 12700K. Alternative is to use Process Lasso extensively to ensure none of your important tasks end up on E-cores, which is a hassle.
 
  • Like
Reactions: dullard

ryanjagtap

Member
Sep 25, 2021
108
127
96
Or disable E-cores if you go with 12700K. Alternative is to use Process Lasso extensively to ensure none of your important tasks end up on E-cores, which is a hassle.
So what you and @dullard are trying to say is, for users having multiple background apps/programs running it's either go all in with 8P+8E cores or don't get E cores at all, so that thread director doesn't mess up the background tasks. Seems to me that Intel and Microsoft still have to do some optimizations to the Thread Director.
 
Jul 27, 2020
16,312
10,335
106
Seems to me that Intel and Microsoft still have to do some optimizations to the Thread Director.
It wasn't the right solution for power users. The common user won't notice a little slowdown if the Thread Director messes up but the power user will get super annoyed. They need to give some control of the Thread Director to the end user through some tool so it can be tweaked to make the correct core choices.
 
  • Like
Reactions: ryanjagtap

dullard

Elite Member
May 21, 2001
25,066
3,415
126
So what you and @dullard are trying to say is, for users having multiple background apps/programs running it's either go all in with 8P+8E cores or don't get E cores at all, so that thread director doesn't mess up the background tasks. Seems to me that Intel and Microsoft still have to do some optimizations to the Thread Director.
The Thread Director is doing what Intel intended: to give a snappy high performance to the program that the user is actively using. For most people, that is the right decision. But, Anandtech posters are NOT the same as most people. We tend to have much more exacting demands on our computers. For those people, yes, go 8+E cores or none at all. There are work arounds (such as putting your important background task on another monitor or set Windows to High-Performance mode instead of the default Balanced Power Plan), but like igor_kavinski said, they are all a hassle.

4 E cores is just a horrible situation to be in for a computer enthusiast. Most reviewers complained about it.
 
Last edited:

Justinbaileyman

Golden Member
Aug 17, 2013
1,980
249
106
That was bad HOW exactly? It was their mistake. You could have sold the laptop. They needed to learn. Now they will continue to mix up orders since they expect customers to return them at their own expense.
That is the same thing my wife told me to do. I just cant do that kind of thing, to me that would be stealing and I couldn't live with my self knowing I did something like that. Like it would literally eat at me and I would end up having nightmare's and get all paranoid and stuff , besides I have horrible luck as it is in life and I dont need to add to it.
I mean I don't believe in like karma or anything because I am a Christian and all, but I have seen and gone through some really crazy stuff you wouldn't believe that makes me think something like Karma could be real.
 
  • Like
Reactions: igor_kavinski

Justinbaileyman

Golden Member
Aug 17, 2013
1,980
249
106
Sorry don't know if my post was against the rules, I was jus trying to answer your question. Not trying to point fingers or judge anyone. Just explaining my own crazy thoughts and idea's not that I always make sense.
 

Justinbaileyman

Golden Member
Aug 17, 2013
1,980
249
106
Anyways Back on topic here guys. What about using linux with Alderlake?? Is the Scheduler better in linux then Windows 11 or doesn't anyone use linux with Alderlake yet?? I would really like to see a windows vs linux for Kdenlive and say Handbrake using any Alderake cpu.