Question Alder Lake - Official Thread

Page 105 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

LikeLinus

Lifer
Jul 25, 2001
11,518
670
126
@Markfw

Why are you spending $800/mo to run so many systems? Wouldn't it be cheaper to put them in a COLO and use their power at a reduced cost?
Because Intel is the devil!!! Doesn't matter how many watts Mark consume and the 1000+ cores he use. He wont mention that, but anyone else using anything he doesn't approve of, we are all trash energy consumers and wasteful humans! But, don't worry, he can use 10k watts and our 200 watts make us a POS.
 

LikeLinus

Lifer
Jul 25, 2001
11,518
670
126
Many here think the 5800X3D may very well beat the 12900KS, not just me. I am a realist, not a blinded Intel fan.

Biased much ? Do you own both ? No. Do I ? Yes
Look at the posts below yours. Many here think the 5800X3D may very well beat the 12900KS, not just me. I am a realist, not a blinded Intel fan.

Biased much ? Do you own both ? No. Do I ? Yes ! And the power draw ?? Everybody knows thats the 12900KS will be a power hog and hot. Its right in the techpowerup title. Best all round ? Show me a review that says that. Everything I read says mixed bag, and thats before the 5800X3D for gaming. I admit right now ADL is best for gaming, but not productivity, and that aligns with most reviews.
Wait, you own a 5800X3d??
 
  • Haha
Reactions: Zucker2k

Hulk

Diamond Member
Oct 9, 1999
4,225
2,015
136
With some of the previous replies, "hobbies" and the 12900k and ks come into "on-topic" IMO.

With that said, one of my hobbies started out that my house was a lan party place for my son and all his friends who could not afford nice computers. When they were idle, I started playing in the DC arena. When cancer hit me (and by then the kids had moved on in life), the DC work turned into a hobby. But with all these computers working to help cure cancer, the electric bill started to become an issue, hence my reluctance in buying inefficient processors that use too much electricity. $800 a month is not a pittance.

Back to the 12900ks.... It is the most wretchedly inefficient processor Intel has possibly ever made. But if someone wants the best in gaming, and does not care about inefficiency (like Adamk47) Thats fine, each to his own. They just need to respect others opinions. But that is a real niche product. Not sure if the 5800X3D will be since it is efficient, and much more affordable Almost 1/2 the price of the 12900ks. And I think that myself and Adam are the two people that best fit your first sentence.

Edit: And on-topic again, the 12900k is 15% lower vs the 5950x in multi-threaded benchmarks according to this, and takes almost twice the power to do it !

That chart uses the 3990x as the highest rated processor, and the rest are a percentage of that one, 5950x=53%, 12900k=38%

I've been here for a while. It seems like you've had some hard knocks but you continue on while doing your best to help others, that's inspirational. Your bottom line with CPU's is how you can use them to help others. Pretty damn cool and hard to argue with actually.
 
  • Love
  • Like
Reactions: scannall and Markfw

LikeLinus

Lifer
Jul 25, 2001
11,518
670
126
Guys, Mark has had cancer. Give him a break and let him be.
Wow, wait, because he has cancer, he is unable to be a corporate shill??? I have screenshots of him saying he has unreleased AMD CPUs. Look, I got ban once because I accused Rollo of being a shill, but now I have screen shots of a member who says they have CPUS that are not released to the public and that same person is shaming the competition and providing recommendations for said CPU?




This is too much and too far.
You are implying a member is a corporate shill.
You are calling out the mods over a rollo incident
You are harassing a member while in the process of derailing the thread.


esquared
Anandtech Forum Director
 
Last edited by a moderator:
  • Like
Reactions: Ajay and SMU_Pony

DrMrLordX

Lifer
Apr 27, 2000
21,629
10,841
136
So, you hate Ferrari's, Lambos, Porche and every sports car ever made.

To be fair, a disproportionate number of those cars (particularly Ferraris and Lambos) are impractical and awful to drive and own. They look great and go pretty fast. Even the 12900KS isn't as awkward to purchase and use as some Ferraris or Lambos.

Because Intel is the devil!!!

If Mark sold all his own hardware and rented servers at a colocation service, he'd probably wind up with a bunch of Milan since that happens to be some of the best/most efficient silicon available for DC workloads. It's not like him taking @Tech Junky 's suggestion would put him on a bunch of Cascade Lake-SP or Ice Lake-SP.
 

LikeLinus

Lifer
Jul 25, 2001
11,518
670
126
He meant that he has both Zen 3 and ADL CPUs.
His exact quote says "
Look at the posts below yours. Many here think the 5800X3D may very well beat the 12900KS, not just me. I am a realist, not a blinded Intel fan. Biased much ? Do you own both ? No. Do I ? Yes ! "

I have the entire conversation screen captured. He changed it and it's no longer there.
 

LikeLinus

Lifer
Jul 25, 2001
11,518
670
126
To be fair, a disproportionate number of those cars (particularly Ferraris and Lambos) are impractical and awful to drive and own. They look great and go pretty fast. Even the 12900KS isn't as awkward to purchase and use as some Ferraris or Lambos.



If Mark sold all his own hardware and rented servers at a colocation service, he'd probably wind up with a bunch of Milan since that happens to be some of the best/most efficient silicon available for DC workloads. It's not like him taking @Tech Junky 's suggestion would put him on a bunch of Cascade Lake-SP or Ice Lake-SP.
I know what I saw and what he said, if even in mistake.
 

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,551
14,510
136
To be fair, a disproportionate number of those cars (particularly Ferraris and Lambos) are impractical and awful to drive and own. They look great and go pretty fast. Even the 12900KS isn't as awkward to purchase and use as some Ferraris or Lambos.



If Mark sold all his own hardware and rented servers at a colocation service, he'd probably wind up with a bunch of Milan since that happens to be some of the best/most efficient silicon available for DC workloads. It's not like him taking @Tech Junky 's suggestion would put him on a bunch of Cascade Lake-SP or Ice Lake-SP.
I got all the hardware starting with being a lan party house. It evolved. And its fun to have a farm to play with. I live alone, as the son is gone and married, and all this hardware to play with. Renting a COLO would be boring....
 

Tech Junky

Diamond Member
Jan 27, 2022
3,410
1,144
106
Renting a COLO would be boring
How so? You still get to play with everything... it's not like you're ripping apart everything daily and tweaking things. Get it off premise and save some cash for the next objects of your affection to play with while they do the grunt work in a cage somewhere.
 

epsilon84

Golden Member
Aug 29, 2010
1,142
927
136
I think if you go up in this thread, you will find where I post that the P-cores are stronger than the 5950x's, but there are only 8 of them, so anything over 8C/16T will get worse in proportion to the 5950s. So, yes, in some lightly threaded apps, its beats the 5950x. But when all cores are loaded (being 24 for 12900k or 32 for 5950x), the 5950x pulls ahead. 8 strong+8 weak <>16 pretty strong. It depends on the load, I proved it for myself in DC apps. There is a thread in the DC forum where I talk about this.
Also, no mention of power usage ? Let me read more....

So, yes, I will not argue, since I already said this ! (in effect)

Edit: I read the whole thing. No mention of power usage, but the 12900k is 5.5% faster per this, but at 70% more power ???? (just by the 142 vs 241 watt, no real data from the site)

It's an aggregate of benchmark scores, not a review.

I posted that mostly in response to you claiming the 5950X wins in most benchmarks except gaming, which is factually wrong.

The link specifically shows that from 405 benchmarks, the 12900K wins 55%.

Yes, relatively speaking the 5950X does better in MT workloads, but real world usage is often bursty and not 100% parallel. Your DC usage is an exception, not a rule. The vast majority of people don't have their CPUs pegged at 100% usage 24/7.
 
  • Love
Reactions: Zucker2k

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,551
14,510
136
How so? You still get to play with everything... it's not like you're ripping apart everything daily and tweaking things. Get it off premise and save some cash for the next objects of your affection to play with while they do the grunt work in a cage somewhere.
I get to choose what hardware, then benchmark, compare, tweak..... My video cards are running at 67% power !
 
  • Like
Reactions: Drazick

Tech Junky

Diamond Member
Jan 27, 2022
3,410
1,144
106
I get to choose what hardware, then benchmark, compare, tweak..... My video cards are running at 67% power !
You can still do all of that remotely. The only thing you couldn't do is touch the HW unless you go there or pay for smart hands to do it for you if it's already on site.
 

LikeLinus

Lifer
Jul 25, 2001
11,518
670
126
I get to choose what hardware, then benchmark, compare, tweak..... My video cards are running at 67% power !
Curious, because I have video and screen shots. You said:

Look at the posts below yours. Many here think the 5800X3D may very well beat the 12900KS, not just me. I am a realist, not a blinded Intel fan. Biased much ? Do you own both ? No. Do I ? Yes ! "

Are you a realist having an AMD cpu that isn't released vs an Intel CPU that is released?
 

DrMrLordX

Lifer
Apr 27, 2000
21,629
10,841
136
The link specifically shows that from 405 benchmarks, the 12900K wins 55%.

To be honest, let's not pretend like that's a huge win for the 12900K. It's squaring off against a CPU that was a year old when the 12900K came out, and not winning everything. I think Mark does exaggerate the situation a bit, though I will also point out that most of those reviews do not focus on fixed power limits. @142W the 5950X probably does win a majority of MT benchmarks over the 12900K.
 
  • Love
Reactions: Markfw

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,551
14,510
136
It's an aggregate of benchmark scores, not a review.

I posted that mostly in response to you claiming the 5950X wins in most benchmarks except gaming, which is factually wrong.

The link specifically shows that from 405 benchmarks, the 12900K wins 55%.

Yes, relatively speaking the 5950X does better in MT workloads, but real world usage is often bursty and not 100% parallel. Your DC usage is an exception, not a rule. The vast majority of people don't have their CPUs pegged at 100% usage 24/7.
It appears that is from user submitted tests, but I won't dispute it on average, however, its not a true review or real. Also, again, no power statistics. And winning 55% where not all workloads are 100% loaded, but loosing 45% kind of makes sense.

Edit: To add.... For your usual user, a 12400 (as igor posted) for like $210 ? Thats a great CPU for that price. 6 P-cores and a 65 watt PL1 ? (what is PL2). But for a power user, with all the cores(and most likely multiple machines), power usage can be a big deal, thats why I keep mentioning power usage.
 
Last edited:
  • Like
Reactions: Drazick

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,551
14,510
136
To be honest, let's not pretend like that's a huge win for the 12900K. It's squaring off against a CPU that was a year old when the 12900K came out, and not winning everything. I think Mark does exaggerate the situation a bit, though I will also point out that most of those reviews do not focus on fixed power limits. @142W the 5950X probably does win a majority of MT benchmarks over the 12900K.
Yes, exactly. And @Tech Junky 's point. I think the new hybrid stuff is not revealed correctly in all OS's. I have been told that linux need 5.16 to utilize the P/E cores correctly, but its not available in a publicly approved OS yet.

Edit: my 12700F is dual boot, win 11 and linux. But since linux is at least 20% faster in most things, I am using that boot and 5.13 kernal.
 
  • Like
Reactions: Drazick

Tech Junky

Diamond Member
Jan 27, 2022
3,410
1,144
106
I updated to 5.17 last week and it's working well. Now, if things are being measured correctly or not when testing is a different story. If some indicators are that using 5.18 corrects everything as designed then that's coming soon as they just released rc1.

Personally 5.17 was painless compared to jumping through hoops between 5.15 - 5.16 versions. It seems like a better / complete release that factors HW in more comprehensively than prior releases. @Markfw I left you some info on how to on another thread you had going but, here's the link again anyway for the kernel.


With things changing so quickly I normally update every Monday to keep things in order / patched. Relying on the OS to do the updates is pointless as they're usually 2-3 major releases behind the published release schedule.
 

carrotmania

Member
Oct 3, 2020
66
184
76
So, is the new AMD CPU an act of desperation? I mean, why are they trying to one up a company they "supposedly" beat and have dominated? Let that sink in. Insert excuses here...

VCache chips and associated generalised preformance were announced ~6 months before Alderlake chips hit the market. And it was originally "shown" on a 5900, not a 5800. And I'm sure Intel have managed to get at least some hard info, due to milanX shipping for some months now, to MS at the very least.

If you think desperation means AMD announcing new product months in advance of hints about the competition, and then scaling back on that announcement, then we'll have to agree to call every new product, by every company ever, desperation.
 
  • Like
Reactions: scannall

Markfw

Moderator Emeritus, Elite Member
May 16, 2002
25,551
14,510
136
I updated to 5.17 last week and it's working well. Now, if things are being measured correctly or not when testing is a different story. If some indicators are that using 5.18 corrects everything as designed then that's coming soon as they just released rc1.

Personally 5.17 was painless compared to jumping through hoops between 5.15 - 5.16 versions. It seems like a better / complete release that factors HW in more comprehensively than prior releases. @Markfw I left you some info on how to on another thread you had going but, here's the link again anyway for the kernel.


With things changing so quickly I normally update every Monday to keep things in order / patched. Relying on the OS to do the updates is pointless as they're usually 2-3 major releases behind the published release schedule.
When you reply to my PM (so I can find it next time with the full procedure) I will let you know what I find, how it works. As some have said, my feedback is on FULL cores being utilized, but that is alderlakes greatest weakness right now, aside from power consumption.
 
  • Like
Reactions: Drazick

Tech Junky

Diamond Member
Jan 27, 2022
3,410
1,144
106
When you reply to my PM (so I can find it next time with the full procedure) I will let you know what I find, how it works. As some have said, my feedback is on FULL cores being utilized, but that is alderlakes greatest weakness right now, aside from power consumption.
Tweaking things between only P and combination or only 1P + all E is outside the param's for how it works as a whole.

Different monitoring programs though give different readings.

Glances::
1649039142624.png

Sensors:
1649039225317.png

Webmin:
1649039310653.png