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.

 

[JoeRambo]

Not the bloodbath you anticipated. I agree with the assessment that gaming numbers could be better with low latency DDR4, on both sides, but I'm more intrigued with ADL-S DDR4 vs DDR5 comparison.

Yeah, it's not as bad, Anandtech won't be so generuos with their JEDEC timings, but then again ZEN3 won't be running 3600C16 either, so i guess pecking order will still be the same.
OC, memory OC, disabled E cores, DDR4 vs DDR5, ton of things are missing.

 

[majord]

T
* 12600k is 30% faster than 5600x and only $10 more expensive, representing a generational leap over the 5600x
* 12600k never went above 125w, and beats 5600x by more percentage points in performance than power consumption.

Last line confirms what I said yesterday about ADL-S efficiency compared to Zen 3, when the chip is not being pushed way beyond power limits. Still, this is only one review so let's wait for the rest. Interesting few hours ahead.

5600x draws ~75w in their blender test. That's 50-60% more power for 30% more perf

 

[Zucker2k]

I disagree; the 12900K is only 10% faster than the 5900X in productivity, and in gaming the 12600K is hot on it's heels for half the price. If you're mainly gaming, then it makes no sense to go for the 12900K; if you're into productivity the 5900X is better value.

Just 3% behind 5950x in productivity on average, while being faster in everything else less threaded and gaming? I don't think so at all.

Looks a bit underwhelming to me tbh. Gaming not by that much ahead (less than I expected), Zen 3 with V-Cache should be enough for AMD to take the "gaming crown" back.

As for productivity, well. Performance by it's own is not that bad, but power consumption though.

This assumes ADL isn't extracting all the performance from current gaming cards. Moreover, gaming doesn't scale linearly, and Zen 3D is only going to be faster in certain titles. I won't get my hopes up too much.

 

[coercitiv]

* 12600k never went above 125w, and beats 5600x by more percentage points in performance than power consumption.

Linus has the 12600K leading by 40-45% in rendering, and the power consumption at 125W is ~65% higher than the 75W of 5600X in the Blender test.

Last line confirms what I said yesterday about ADL-S efficiency compared to Zen 3

We can argue than lowering power consumption on 12600K will result in lower relative performance drop (meaning 12600K may still end up close in efficiency with 5600X at ISO power), but then again we're still talking 6+4 cores versus 6 cores, so I don't understand how your comment from yesterday regarding Golden Cove holds up. If anything it seems this early review confirms the opposite.

 

[Carfax83]

I disagree; the 12900K is only 10% faster than the 5900X in productivity, and in gaming the 12600K is hot on it's heels for half the price. If you're mainly gaming, then it makes no sense to go for the 12900K; if you're into productivity the 5900X is better value.

I don't disagree that the 5900x is a better value than the 12900K for productivity. I was speaking purely from a performance perspective. The 12900K isn't much further behind than the 5950x despite having significantly less threads and performance cores.

Those efficiency cores help out undoubtedly but they are still apparently fairly weak for heavy workloads, and it's obvious that the performance cores have to be clocked very high to catch up with the 5950x, causing the power consumption to skyrocket.

Future efficiency cores with higher IPC, greater counts and wider vectors should be able to remedy that vulnerability and lead to lower power consumption overall.

 

[DrMrLordX]

Setting power above 125 W is for two types of people:
1) Those who want performance above all.
2) People who want to say the chip has terrible performance per watt.

That is why good reviews tend to show things at stock (125W) and when pushed to extremes.

Sadly, 125W isn't stock for the 12900k. On a lot of boards, it's going to pull as much power as possible until:

1). Thermal throttling of cores or
2). Thermal throttling of VRMs or
3). Socket power limits or
4). Boost clock limits

are achieved.

I disagree; the 12900K is only 10% faster than the 5900X in productivity, and in gaming the 12600K is hot on it's heels for half the price. If you're mainly gaming, then it makes no sense to go for the 12900K; if you're into productivity the 5900X is better value.

We need to see some more reviews, but it does look like it's shaping up that way.

 

[coercitiv]

The best news so far is gaming power consumption. Productivity is a mixed bag in terms of efficiency, but at least the performance is there, and the pricing is aggressive enough.

 

[tamz_msc]

Just 3% behind 5950x in productivity on average, while being faster in everything else less threaded and gaming? I don't think so at all.

The 5950X is 6% faster than the 12900K in productivity; while that's impressive on its own, the pricing means that the 5900X is a whopping $100-125 cheaper than the 12900KF/K going by Newegg pricing.

In terms of perf/$ in productivity, the math is pretty clear - the 5900X is a clear winner.

 

[TheELF]

Sadly, 125W isn't stock for the 12900k. On a lot of boards, it's going to pull as much power as possible until:

CPU stock and mobo stock is not the same, you could also have said that a lot of overclockers will do these things, doesn't make it stock though.
125W is the stock efficiency setting for the 12900k, basta.

 

[Zucker2k]

The 5950X is 6% faster than the 12900K in productivity; while that's impressive on its own, the pricing means that the 5900X is a whopping $100-125 cheaper than the 12900KF/K going by Newegg pricing.

In terms of perf/$ in productivity, the math is pretty clear - the 5900X is a clear winner.

So you're going to leave everything else out and concentrate on productivity, and the dollar value of 10%? Okay? Hehe

 

[TheELF]

The 5950X is 6% faster than the 12900K in productivity; while that's impressive on its own, the pricing means that the 5900X is a whopping $100-125 cheaper than the 12900KF/K going by Newegg pricing.

In terms of perf/$ in productivity, the math is pretty clear - the 5900X is a clear winner.

It's also a good bit slower.

 

[tamz_msc]

So you're going to leave everything else out and concentrate on productivity, and the dollar value of 10%? Okay? Hehe

There isn't anything else to say about the i9. Gaming wise, the 12600K equals it. Productivity wise, the 5900X is only 10% behind while being significantly cheaper. It's neither here nor there, and it's only going to be of interest to the extreme OC public.

 

[insertcarehere]

The 5950X is 6% faster than the 12900K in productivity; while that's impressive on its own, the pricing means that the 5900X is a whopping $100-125 cheaper than the 12900KF/K going by Newegg pricing.

In terms of perf/$ in productivity, the math is pretty clear - the 5900X is a clear winner.

If we're gonna bag on the 12900k/f (an iGPU has legit value in 2021) for looking bad in terms of perf/$ in productivity vs the 5900X, it should also be pointed out that the same performance metrics used to justify this statement make the 5950X look even worse from a perf/$ perspective. Did people use that as justification to make the 5950x seem "unimpressive"?

 

[tamz_msc]

It's also a good bit slower.

I'll take 10% slower performance for 17% cheaper price any day. And that's only accounting for the CPU price alone.

 

[Head1985]

Alder lake looks good.Gaming power consuption is good.Even better than ryzens.12600KF destroys 5600x in everything and is also cheaper.Cant wait for 12400F, 12700F.AMD need massively reduce prices.Win for consumers competition is back.

 

[Zucker2k]

I'll take 10% slower performance for 17% cheaper price any day. And that's only accounting for the CPU price alone.

Keep that thought. 12700k incoming!

 

[tamz_msc]

If we're gonna bag on the 12900k for looking bad in terms of perf/$ in productivity vs the 5900X, it should also be pointed out that the same performance metrics used to justify this statement make the 5950X look even worse from a perf/$ perspective. Did people use that as justification to make the 5950x seem "unimpressive"?

Halo products are always unimpressive, at least I consider them so.

 

[TheELF]

There isn't anything else to say about the i9. Gaming wise, the 12600K equals it. Productivity wise, the 5900X is only 10% behind while being significantly cheaper. It's neither here nor there, and it's only going to be of interest to the extreme OC public.

I'll take 10% slower performance for 17% cheaper price any day. And that's only accounting for the CPU price alone.

Yeah and compared to the 5950x the 12900k is less than 10% slower and is $200 cheaper.
Intel found the perfect balance to place this CPU for maximum impact, the 5950x which is AMDs highest margin CPU is much less desirable now.
The 5900x looks much better against it but then you get into 12600k territory, we have to see how these two will compare.
I'm not disagreeing with you, anybody can choose whichever CPU is best for them, there are always alternatives.

 

[tamz_msc]

Keep that thought. 12700k incoming!

Yeah It'll make the i9 look even more silly.

 

[Head1985]

More like 12400F and 12700F incoming.Those will be cpus to buy.

 

[Timorous]

Just did a price comp at OCUK.

12600K + Z690 Tomahawk + 32GB 5200C38 ram is £850.
5600X + X570 Tomahawk + 32GB 3600C16 ram is £610.

The 12600K is good value, the platform as a whole is not for just gaming (EDIT: and for just gaming you could easily go B550 and save some money on the 5600X build). If you want productivity you have enough budget to get a 5800X and still pay less or you can pay £80 more and drop in a 5900X.

EDIT: That Z690 is DDR4 and it costs £260. There are £200 DDR5 mobos but no idea if they are actually any good. I await the motherboard testing. If I go GB Gaming X vs Gaming X then the price difference drops by £50. not enough to make the 12600K build cheaper than a 5800X build but it does make the 5900X quite a stretch.

 

[Head1985]

Just did a price comp at OCUK.

12600K + Z690 Tomahawk + 32GB 5200C38 ram is £850.
5600X + X570 Tomahawk + 32GB 3600C16 ram is £610.

The 12600K is good value, the platform as a whole is not for just gaming (EDIT: and for just gaming you could easily go B550 and save some money on the 5600X build). If you want productivity you have enough budget to get a 5800X and still pay less or you can pay £80 more and drop in a 5900X.

You dont need DDR5 for 12600K and if you comparing ryzen without igpu then compare it to intel without GPU and thats 12600KF.This driving me nuts.People always comparing ryzen without igpus to 12600K/12700K/12900K with igpus.
And if you want ultimate bank for buck wait for non K cpus to drop in January. 12700F should cost around 300USD and 12400F bellow 200USD.

 

[coercitiv]

This driving me nuts.

Take it easy then. We're just getting started.

 

[arandomguy]

The best news so far is gaming power consumption.

Gaming power consuption is good.

Gaming consumption for Intel has been widely divergent from so called "productivity" power consumption ever since Intel started clocking closer to the limit out of the box. It's not just a ST vs. MT load factor difference either.

Due to divergent methods of dynamic clocking along with differing design targets in silicon/manufacturing with AMD's Zen this has since resulted a disconnect in terms of how power consumption actually compares depending on the task load, as a task like encoding/rendering is very different than gaming.

I've ranted about this issue before in terms of how tech reviewers for the most part solely convey power consumption numbers via a task such as rendering/encoding or similar and therefore rely on that as a measure solely for efficiency. This has created a very muddled narrative in terms of power consumption/efficiency with the public, especially as their performance content and target audience leans much more so (if not entirely) into gaming (or at least non rendering/encoding/etc. tasks).

Reviewers really should have power consumption data in idle, "desktop" (eg. browsing, video watching, etc.) gaming, productivity burst, and productivity throughput scenarios, as opposed to just the last one.

 

[TheELF]

Gaming consumption for Intel has been widely divergent from so called "productivity" power consumption ever since Intel started clocking closer to the limit out of the box.

Due to divergent methods of dynamic clocking along with differing design targets in silicon/manufacturing with AMD's Zen this has since resulted a disconnect in terms of how power consumption actually compares depending on the task load, as a task like encoding/rendering is very different than gaming.

I've ranted about this issue before in terms of how tech reviewers for the most part solely convey power consumption numbers via a task such as rendering/encoding or similar and therefore rely on that as a measure solely for efficiency. This has created a very muddled narrative in terms of power consumption/efficiency with the public, especially as their performance content and target audience leans much more so (if not entirely) into gaming (or at least non rendering/encoding/etc. tasks).

Reviewers really should have power consumption data in idle, "desktop" (eg. browsing, video watching, etc.) gaming, productivity burst, and productivity throughput scenarios, as opposed to just the last one.

Also they only show maximum peak instead of average and not at stock settings but with power limits lifted, if they aren't only showing overclocking numbers.

Intel Core i7-11700K Review: The Chip of Last Resort

High power consumption meets high pricing, again.

www.tomshardware.com