Question Alder Lake - Official Thread

[Hulk]

With the release of Alder Lake less than a week away and the "Lakes" thread having turned into a nightmare to navigate I thought it might be a good time to start a discussion thread solely for Alder Lake.

 

[Markfw]

I actually agree with you that a cross platform upgrade doesn't make the most sense since there are upgrade options available on the AM4 side, but in what world is a 5800X to 12700K a 'side grade'?

You get better IPC + higher clocks with the 12700K, plus more cores/threads. The 12700K wins in every metric except for worse power consumption, and even then its not in 12900K territory because Intel didn't squeeze it to an inch of its life. You should know this since you own a 12700K yourself.

The 12700K is a 5900X competitor, the 5800X is in the 12600K class.

On paper, its looks good, and in gaming it wins. Not in all things. For example, (DC work) in primgrid, 2 different projects. One it gets creamed in my even a 3950x, and another, it beats the 5950x (well 8 cores of a 5950x). So it really depends on the exact workload. An power is all over the place.275 to 161 (at the wall, no video usage).

In general, I won't disagree, the 5900x is a closer competitor. He looked like a gaming person, hence the 5800X3D recommendation. As far as the e-cores being used, even you said there are outliers. Just seems too early to trust it. I had to disable them to get good performance (but I am on unix primarily.)

 

[LikeLinus]

I actually agree with you that a cross platform upgrade doesn't make the most sense since there are upgrade options available on the AM4 side, but in what world is a 5800X to 12700K a 'side grade'?

You get better IPC + higher clocks with the 12700K, plus more cores/threads. The 12700K wins in every metric except for worse power consumption, and even then its not in 12900K territory because Intel didn't squeeze it to an inch of its life. You should know this since you own a 12700K yourself.

The 12700K is a 5900X competitor, the 5800X is in the 12600K class.

FYI: You should seek real reviews and performance charts. There are certain people here who are going to recommend said product, no matter what. You aren't going to get an unbiased opinion, period. Just look at the benchmarks and the people who can actually test the products and know what they are talking about, then make your decision. Ultimately, most "users" don't really have the resources or testing ability to form a valid opinion. Look to the people who actually know what they are doing and conduct real test and not in just one specific area. Look for overall performance.

 

[Markfw]

FYI: You should seek real reviews and performance charts. There are certain people here who are going to recommend said product, no matter what. You aren't going to get an unbiased opinion, period. Just look at the benchmarks and the people who can actually test the products and know what they are talking about, then make your decision. Ultimately, most "users" don't really have the resources or testing ability to form a valid opinion. Look to the people who actually know what they are doing and conduct real test and not in just one specific area. Look for overall performance.

And don't forget power efficiency. Overall performance really doesn't matter when its almost impossible to cool, or your electric bill goes through the roof.

 

[ondma]

FYI: You should seek real reviews and performance charts. There are certain people here who are going to recommend said product, no matter what. You aren't going to get an unbiased opinion, period. Just look at the benchmarks and the people who can actually test the products and know what they are talking about, then make your decision. Ultimately, most "users" don't really have the resources or testing ability to form a valid opinion. Look to the people who actually know what they are doing and conduct real test and not in just one specific area. Look for overall performance.

Overall performance would probably favor the 12700k *more* than simply gaming performance though. Both are 8 cores, but the 12700k also has 4 e-cores that would help in productivity. Gaming probably not so much, unless one is running background tasks which are shifted to the e cores. I could see a rationale for selecting the 12700k over the 5800x for a build from scratch, but for an upgrade, I dont see it making any sense unless there is some specific game/task that runs unusually well on Intel.

Also, might want to wait for the 5800 3d stacked cache chip (dont recall exactly what it is called), and see how that performs relative the the 5800x and 12700k. Price (and availability) most likely will be sketchy, but it would not require a new motherboard.

 

[LikeLinus]

And don't forget power efficiency. Overall performance really doesn't matter when its almost impossible to cool, or your electric bill goes through the roof.

I get that is your go to quote. But this is coming from someone who runs their PCs 24/7 at max performance for DC. Which is a VERY VERY limited segment,.

Let's quote your own signature.

All Doing Rosetta or WCG + F@H- -2 x Ryzen 3900x \ Ryzen 3950x \ 5 x Ryzen 5950x \ 12700F
EPYC 7401 \ Dual EPYC 7601 (64c/128t for both) \ EPYC 7742 (64c/128t) \ 7551 EPYC 32c/64t
2 x (Threadripper 1950X) EPYC 7452 \ Xeon E5-2683v3 \ EPYC 7742 (64c/128t)
TR 2970wx \ TR 2990WX \ TR 2990WX \ EPYC 7B12 @2.6 ghz
3070, 3080TI, 3070TI, 6 x (2080TI) 3 x (2060) 2 x (1080TI) , 1070TI and 2060 super, 1060

So do you REALLY think that a single CPU used for X amount of hours a day is going to push their energy bill? I mean, really? It's just an excuse. Funny enough, I never see you mention the power consumption used by people doing DC and how they run their computers so hard and full time? (Are you exempt to your own preaching?) Is that better or worse than someone running a single CPU in their rig for enjoyment or work? Do you think you have a lower energy use footprint than someone asking for a PC recommendation?

If we want to be REAL about it.....

There is a minimal impact to power usage, cost and such. I was running a 9900K before my 5950x. I noticed no real difference in my power bill or heating/cooling of the house. I mean, it's a PC sitting in my office. I, of course, have far more PC and servers, but you always want to represent the worst case, which isn't what most users are asking about. If you want to provide unbiased absolute recommendations, look past the whole DC and personal usage. The power/heat difference looks great on paper, but what real effect does it have in real world?

 

[Markfw]

I get that is your go to quote. But this is coming from someone who runs their PCs 24/7 at max performance for DC. Which is a VERY VERY limited segment,.

Let's quote your own signature.

All Doing Rosetta or WCG + F@H- -2 x Ryzen 3900x \ Ryzen 3950x \ 5 x Ryzen 5950x \ 12700F
EPYC 7401 \ Dual EPYC 7601 (64c/128t for both) \ EPYC 7742 (64c/128t) \ 7551 EPYC 32c/64t
2 x (Threadripper 1950X) EPYC 7452 \ Xeon E5-2683v3 \ EPYC 7742 (64c/128t)
TR 2970wx \ TR 2990WX \ TR 2990WX \ EPYC 7B12 @2.6 ghz
3070, 3080TI, 3070TI, 6 x (2080TI) 3 x (2060) 2 x (1080TI) , 1070TI and 2060 super, 1060

So do you REALLY think that a single CPU used for X amount of hours a day is going to push their energy bill? I mean, really? It's just an excuse. Funny enough, I never see you mention the power consumption used by people doing DC and how they run their computers so hard and full time? (Are you exempt to your own preaching?) Is that better or worse than someone running a single CPU in their rig for enjoyment or work? Do you think you have a lower energy use footprint than someone asking for a PC recommendation?

If we want to be REAL about it.....

There is a minimal impact to power usage, cost and such. I was running a 9900K before my 5950x. I noticed no real difference in my power bill or heating/cooling of the house. I mean, it's a PC sitting in my office. I, of course, have far more PC and servers, but you always want to represent the worst case, which isn't what most users are asking about. If you want to provide unbiased absolute recommendations, look past the whole DC and personal usage. The power/heat difference looks great on paper, but what real effect does it have in real world?

First, digging on someone trying to help cure cancer will not buy you any good will. Even though I greatly dislike some people, after having had cancer (and may still) I would not wish it on anyone.

Second, notice I said cooling and power. My 12700F will hit 300 watts sometimes, then quickly throttle down to 161 when AVX-512 kicks in. If I had a 12900k, I imagine at the power curve it would be much harder to cool, and thats been said by all sorts of reviewers. I think its well known that Intel put the power curve and wattage limits so high on the 12900k, just to win a few extra benchmarks. The 12700F is much better, but still at times can be trying.

 

[Exist50]

And don't forget power efficiency. Overall performance really doesn't matter when its almost impossible to cool, or your electric bill goes through the roof.

Power consumption is broadly comparable under gaming workloads, which do not max out the processor.

 

[biostud]

In gaming scenarios Alder Lake and zen3 uses roughly the same amount of power, as it doesn't put full load on the CPU. It is first when you do work that loads all cores that intels power consumption skyrockets compared to zen3.

 

[LikeLinus]

First, digging on someone trying to help cure cancer will not buy you any good will. Even though I greatly dislike some people, after having had cancer (and may still) I would not wish it on anyone.

Second, notice I said cooling and power. My 12700F will hit 300 watts sometimes, then quickly throttle down to 161 when AVX-512 kicks in. If I had a 12900k, I imagine at the power curve it would be much harder to cool, and thats been said by all sorts of reviewers. I think its well known that Intel put the power curve and wattage limits so high on the 12900k, just to win a few extra benchmarks. The 12700F is much better, but still at times can be trying.

First, I didn't dig on anyone trying to cure cancer, but who says DC is the right answer? I mean SETI went on for how long, yet it found nothing? I've worked with a huge academic organization and children's hospital on this very subject. Years and years ago, I even conducted a PC donation initiative on AGN/PC Abuses for members to donate PC parts so I could build PCs for Children living in on-site rooms/wings (of the children's hospital and off-site) so they had access to the internet and ability to do things outside of the constant medical care. I provided tax exemption forms for those individuals to write-off those parts off and did this on my own free will.

My wife worked for over a decade in childrens' cancer research (at one of the top 5 Children cancer hospitals) and I've provided my time and resources to when applicable. I worked events, wine tastings, walks, TV donation drives and other such events. My aunt died of cancer. Sorry, but I've lived through this many times, regardless of my past, I don't let it control my future.

Please don't think you know me or my background. My problem is, you are steering people towards specific thing based on your own individual personal situation. While I understand, it's just not fair because you, knowingly or not, do have an agenda. That's specifically why there is a DC forum. But you are using that as a metric to judge products and the heat/wattage consumption that is really not relevant because, overall, at least in the US, is relatively insignificant.

Has DC ever cured any type of cancer? I mean, I know the answer in this, but just curious if there is a real life proof of concept that you can provide that has saved actual lives from cancer. All of that really has NOTHING to do with cancer research when someone ask for a CPU recommendation for gaming or general usage.

Sure, the 12700F may peak at a certain wattage as needed. Are you saying the AMD does not move from its designated wattage usage? You don't seem to mention the difference between what AMD says it will use and what it actually uses. Why is that? Do we all just assume it's perfectly consistent?

It's sort of like saying "My car gets 30MPG, but if I press the gas pedal and drive it hard, it should use the same MPG!" Sorry, it's just not going to happen. Push something and it demands more power. But you often blow off the advantage that it gives Intel in those various gaming test as being "power hungry" and such. But the whole point is, it does what it's supposed to do.

Look, I sympathise with you and your plight, hell, I'd help, if I could. But fact of the matter is, you can't try to tell everyone to buy a certain product based on personal issues or beliefs. That's why I told them to stick to the testing and actual use cases. Period.

 

[ondma]

Yea, I dont understand the virulent bashing of products based on a person's own specific uses/personal feelings. If one dislikes a particular product or corporation, fine, that is their business/personal feeling, whatever. I just dont understand that translating to such emotionally loaded "advice" directed toward other uses who may have totally different uses and priorities. I mean, a PC is one of the few areas (at least before the shortages) where one can spend a relatively small amount of money and get state of the art performance. It is not like spending for a fast/luxurious car, a big house, or a vacation. Just let people buy what they want, and enjoy it. One will get excellent performance with either Intel or AMD these days, unlike the Bulldozer days when AMD both used more power and gave less performance. So Intel may use an extra 50 or so watts in a gaming scenario. Dont forget you have a probably 200 watt dgpu running full throttle, plus monitors, speakers, etc.

 

[Markfw]

Power consumption is broadly comparable under gaming workloads, which do not max out the processor.

Granted that... But I am talking general usage. Until Intel disables AVX-512, and other software like rendering and encoding and such, it does use more power. I even recommended the 12700k to a few gamers, before 5800X3D seemed to have a recent release date.

 

[LikeLinus]

Granted that... But I am talking general usage. Until Intel disables AVX-512, and other software like rendering and encoding and such, it does use more power. I even recommended the 12700k to a few gamers, before 5800X3D seemed to have a recent release date.

Honest question. Have you or any other review done an actual study on "general usage" with end users to see if they can tell the difference between similar Products from company A and B? Benchmarks are terrific for people like us, but if we are talking GENERAL usage, what is the percentage of people that notice those extra 10 frames at 200FPS, or saving a word document milliseconds quicker, running a powerpoint is faster? I've preached this from YEARS ago. Especially when people were saying "Intel is dead, AMD is dominating, I was like "What are you people smoking? No normal person sees those insignificant differences". Where is the use ability study or "one cup has Pepsi, one has Coke, which do you prefer?"

Yet we all rely on YouTube videos and marketing materials and call the winner. I'm not sure how you don't find that odd. I realize in your specific use case of 27,000 cores mining data might be apparent, but how about the normal users?

 

[Exist50]

Granted that... But I am talking general usage. Until Intel disables AVX-512, and other software like rendering and encoding and such, it does use more power. I even recommended the 12700k to a few gamers, before 5800X3D seemed to have a recent release date.

Since when is your niche compute workload anything close to "general usage"? Gaming is quite likely the most intense use case their system will see, and in that context your claims are outright false.

Also, Alder Lake doesn't support AVX512 out of the box...

 

[LikeLinus]

Since when is your niche compute workload anything close to "general usage"? Gaming is quite likely the most intense use case their system will see, and in that context your claims are outright false.

Also, Alder Lake doesn't support AVX512 out of the box...

Per Intel - AVX512 is used for "scientific simulations, financial analytics, artificial intelligence (AI)/deep learning, 3D modeling and analysis, image and audio/video processing, cryptography and data compression" Plus, not sure why we're talking about it because no current AMD CPU supports AVX512? So how is it a positive or negative if Intel does or doesn't, lol?

Out of all these conversations, it appears "scientific simulations" (I'm assuming DC) is the only one that is used in the majority of questions in this forum. Hell, I do 3D Modeling and Audio/Video Processing probably most than anyone here. I guess I should sell my 5950x and 5800x CPUs for newer Intel CPUs? AVX512 really shouldn't be in the conversation for general use, gaming and such.

 

[Markfw]

Since when is your niche compute workload anything close to "general usage"? Gaming is quite likely the most intense use case their system will see, and in that context your claims are outright false.

Also, Alder Lake doesn't support AVX512 out of the box...

When I disabled the 4 e-cores on my 12700F, and ran primegrid, several times the power jumped from 162 to 300 on my kill-a-watt. StefanR5R found (in log files that in fact it was using AVX-512 at least for a few seconds, we do not know how long. So, yes, out of the box mine certainly does.

 

[LikeLinus]

When I disabled the 4 e-cores on my 12700F, and ran primegrid, several times the power jumped from 162 to 300 on my kill-a-watt. StefanR5R found (in log files that in fact it was using AVX-512 at least for a few seconds, we do not know how long. So, yes, out of the box mine certainly does.

Curiosity, please feel free to post screen shots, how long did the computer jump from 162 to 300watts? Would be nice to see the data and or videos of the jump for an extended period of time.

 

[dullard]

When I disabled the 4 e-cores on my 12700F, and ran primegrid, several times the power jumped from 162 to 300 on my kill-a-watt. StefanR5R found (in log files that in fact it was using AVX-512 at least for a few seconds, we do not know how long. So, yes, out of the box mine certainly does.

1) Set your PL2 value lower or set the tau to a short period of time. Then your CPU won't spike so high. Or at least won't spike for any amount of time that matters.
2) Kill-a-Watt is intended for daily average power, not brief power spikes. What is your daily average power when running your DC projects 24/7?
3) Set your motherboard to turn off AVX-512.
4) Too bad you got the 12700F, the video card (even on idle) consumes a significant amount of your power budget.

 

[Markfw]

Curiosity, please feel free to post screen shots, how long did the computer jump from 162 to 300watts? Would be nice to see the data and or videos of the jump for an extended period of time.

With the 4 e-cores it was taking 275 watts. Right now, its doing 162 with them disabled. I know the 4 e-cores should not take 100 watts, I am just telling you what I see, I watch these all the time, they are pretty real-time.

As for how long it was at 300, I only looked a few times, so I will not say if it was 10 seconds or 10 minutes, but certainly not hours. The log file did not say. Here is what lscpu says:

mark@12700F-linux:~$ lscpu
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
Address sizes: 39 bits physical, 48 bits virtual
CPU(s): 16
On-line CPU(s) list: 0-15
Thread(s) per core: 2
Core(s) per socket: 8
Socket(s): 1
NUMA node(s): 1
Vendor ID: GenuineIntel
CPU family: 6
Model: 151
Model name: 12th Gen Intel(R) Core(TM) i7-12700F
Stepping: 2
CPU MHz: 3133.345
CPU max MHz: 6300.0000
CPU min MHz: 800.0000
BogoMIPS: 4224.00
Virtualization: VT-x
L1d cache: 384 KiB
L1i cache: 256 KiB
L2 cache: 10 MiB
L3 cache: 25 MiB
NUMA node0 CPU(s): 0-15
Vulnerability Itlb multihit: Not affected
Vulnerability L1tf: Not affected
Vulnerability Mds: Not affected
Vulnerability Meltdown: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl and seccomp
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; Enhanced IBRS, IBPB conditional, RSB filling
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Not affected
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts
acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art ar
ch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_f
req pni pclmulqdq dtes64 monitor ds_cpl vmx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pc
id sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand la
hf_lm abm 3dnowprefetch cpuid_fault epb cat_l2 invpcid_single cdp_l2 ssbd ibrs ibpb
stibp ibrs_enhanced tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust
bmi1 avx2 smep bmi2 erms invpcid rdt_a avx512f avx512dq rdseed adx smap avx512ifma
clflushopt clwb intel_pt avx512cd sha_ni avx512bw avx512vl xsaveopt xsavec xgetbv1 x
saves split_lock_detect avx_vnni avx512_bf16 dtherm ida arat pln pts hwp hwp_notify
hwp_act_window hwp_epp hwp_pkg_req avx512vbmi umip pku ospke waitpkg avx512_vbmi2 gf
ni vaes vpclmulqdq avx512_vnni avx512_bitalg avx512_vpopcntdq rdpid movdiri movdir64
b fsrm avx512_vp2intersect md_clear serialize arch_lbr avx512_fp16 flush_l1d arch_ca
pabilities
mark@12700F-linux:~$

And here is Stefan's eval showing the 512 usage. This is all I know.

Page 6 - PrimeGrid Challenges 2022

Page 6 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

 

[Markfw]

Oh, is the the text from one of the jobs (to save you linking). Maybe it was on for 35.710 seconds ?????

<core_client_version>7.16.6</core_client_version>
<![CDATA[
<stderr_txt>
BOINC PrimeGrid wrapper 2.02 (Nov 17 2020 23:46:30)
running ../../projects/www.primegrid.com/sllr2_1.1.0_linux64_201114 -v
LLR2 Program - Version 1.1.0, using Gwnum Library Version 29.8
running ../../projects/www.primegrid.com/sllr2_1.1.0_linux64_201114 -oGerbicz=1 -oProofName=proof -oProofCount=128 -oProductName=prod -oPietrzak=1 -oCachePoints=0 -pSavePoints -q3*2^18447616-1 -d -t16 -oDiskWriteTime=1
Gerbicz check is requested, switching to PRP.
Starting probable prime test of 3*2^18447616-1
Using AVX-512 FFT length 1000K, Pass1=640, Pass2=1600, clm=2, 16 threads, a = 3, L2 = 482*299, M = 144118
Compressed 128 points to 7 products. Time : 35.710 sec.
Testing complete.
01:51:21 (4052): called boinc_finish(0)

</stderr_txt>
]]>

And another:

core_client_version>7.16.6</core_client_version>
<![CDATA[
<stderr_txt>
BOINC PrimeGrid wrapper 2.02 (Nov 17 2020 23:46:30)
running ../../projects/www.primegrid.com/sllr2_1.1.0_linux64_201114 -v
LLR2 Program - Version 1.1.0, using Gwnum Library Version 29.8
running ../../projects/www.primegrid.com/sllr2_1.1.0_linux64_201114 -oGerbicz=1 -oProofName=proof -oProofCount=128 -oProductName=prod -oPietrzak=1 -oCachePoints=0 -pSavePoints -q3*2^18447636+1 -d -t16 -oDiskWriteTime=1
Starting Proth prime test of 3*2^18447636+1
Using all-complex AVX-512 FFT length 1000K, Pass1=640, Pass2=1600, clm=2, 16 threads, a = 11, L2 = 482*299, M = 144122
Compressed 128 points to 7 products. Time : 31.253 sec.
Testing complete.
04:08:41 (5442): called boinc_finish(0)

</stderr_txt>
]]>

 

[igor_kavinski]

The performance difference has got to be substantial for them to go to the trouble of writing AVX-512 optimized code.

 

[Markfw]

The performance difference has got to be substantial for them to go to the trouble of writing AVX-512 optimized code.

These run On Xeons as well, in linux. Maybe even a lot of them, I really don't know.

@LikeLinus , running primegrid, not cancer research, but WCG is down until Apr22, running F@H. And as for results, I can't find it now, but Rosetta@home was used to create a drug that KILLED/CURED covid-19 (in the lab only) in 5 doses. Not sure when it will come to fruition, if ever, but maybe it has steered research to help create the vaccines. It certainly is a project in F@H and Rosetta, not sure about WCG.

Interesting link https://foldingathome.org/diseases/infectious-diseases/covid-19/?lng=en

I am currently contributor number 30 in the WORLD for folding@home.
I am number 29 in the world for Rosetta@home, and number 70 in the world for WCG

And you wanted results:

Papers & Results – Folding@home

foldingathome.org

Now lets get back to Alder lake discussion.

 

[LikeLinus]

These run On Xeons as well, in linux. Maybe even a lot of them, I really don't know.

Rosetta@home was used to create a drug that KILLED/CURED covid-19 (in the lab only) in 5 doses. Not sure when it will come to fruition, if ever, but maybe it has steered research to help create the vaccines.

Not sure this even needs a reply. But, please show me a medical research paper on it, a real (AP) article on it. Saying "not sure where it came from..." or any real news network saying there is a possible cure. Again, you never addressed my post and you are only starting new conversations and diverting the attention to original questions.

 

[biostud]

One thing I think often is forgotten when talking value, is the total cost of motherboard, memory and CPU. It's very seldom you buy a CPU without buying the rest.

 

[Markfw]

One thing I think often is forgotten when talking value, is the total cost of motherboard, memory and CPU. It's very seldom you buy a CPU without buying the rest.

Yes, very true. If you want a decent 690 Alderlake board, the cheapest I could find was $220, but AM4, I have a $150 motherboard on my last 5950x. The memory for DDR4 is the same, but if you go DDR5 on Alderlake, that just skyrockets the price, with not that much more performance. Thats why I went DDR4. And 3200 CL14 is the fastest DDR4 you can use, due to gear down. There are cheap Alderlake boards, but I wanted something midstream, so I went 690.

Edit: even on the low end, socket 1700 is $80, and AM4 is $50.

 

[Shmee]

Hey all, please remember to keep this thread on subject about Alder Lake. For other subjects, please comment in another appropriate thread, or make a new one if it does not exist. Thanks!

-Moderator Shmee