Discussion Intel current and future Lakes & Rapids thread

Page 545 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

deasd

Senior member
Dec 31, 2013
513
724
136
It could be as simple as Sandra having the type of workload called - equal parcels: It takes number of threads @ 24 ( for 8+HT +8 small) and runs an equally sized computation on each. It works fine on homogenous architectures, where each thread has same performance, but fails utterly in hybrid architecture.
On hybrid, big cores complete the workload first and what happens next depends on OS:
1) On hybrid unaware OS like W10 => workload is left to keep on running on small cores, while big core is sat Idle => this decision looks stupid, but in fact it is optimal on previuos architectures, to stick workload whereever it is now to reap benefits from hot caches, hot branch predictors and so on.
2) On hybrid aware OS like W11, CPU is hinting OS that it currently small core is running heavy workload that would be better fit on big core and it gets rescheduled there and total "performance" increases cause workload completes sooner.

Of course due to inflexible workload, even (2) approach idles the small cores during that final strech of runtime and total potential does not reach 100% like it does in things like CB.

Sounds like a mess... if true Sandra & other apps alike have to rewrite the code just for big+little? Win11 isn't enough? Now I see that's why 'N+0' i5s more attractive than any other above i7...
 
  • Like
Reactions: Tlh97 and Joe NYC

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
And the Intel chip uses Turbo!

So? That's built into the power spec. PBO is not. Turning on PBO will increase power usage on an AMD 5-series chip vs. what boost clocks will allow @ stock. So if @Makaveli 's CPU is using more than other CPUs of its kind from reviews, it's because he's using PBO.

Regardl,ess, the entire purpose of his post was to show that AIDA64 FPU uses less power than CBR20, which on his system, it does.
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
So? That's built into the power spec. PBO is not. Turning on PBO will increase power usage on an AMD 5-series chip vs. what boost clocks will allow @ stock. So if @Makaveli 's CPU is using more than other CPUs of its kind from reviews, it's because he's using PBO.

Regardl,ess, the entire purpose of his post was to show that AIDA64 FPU uses less power than CBR20, which on his system, it does.
Do you realize PBO = Turbo? Turbo performance doesn't come free, you know?
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Do you realize PBO = Turbo? Turbo performance doesn't come free, you know?

PBO is not "turbo". It is literal overclocking by completely removing power limits. There is no tau or time limit on how long elevated power levels can be maintained in order to stay within a power spec. It's balls-to-the-walls fast-as-you-can-go (when it works properly).
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
PBO is not "turbo". It is literal overclocking by completely removing power limits.
Certainly not all power limits? What happened to the power cap? The 5800x enjoys the same power cap as the 5900x and 5950x? You've been bashing Intel for ages about releasing unrestricted power chips for ages, now all of a sudden AMD also does it?
 

Rigg

Senior member
May 6, 2020
435
863
106
This should be an interesting review cycle. There seems to be a growing list of major variables at play.

-Intel 'recommended' power and turbo duration limits vs whatever default limits the review board uses
-DDR4 vs DD5
-Windows 11 vs 10

The first variable is obviusly a carryover from previous generations but the other 2 are also huge factors. Valid comparrissons to AMD and previous Intel generations will be questionable at best without all 3 of those things accounted for. That's going to add a lot of extra testing for reviewers that will probably be impossible to accomplish for day 1 reviews. Not to mention DDR5 and W11 performance will be a moving target as the OS matures and better DDR5 kits release. If you think this thread is contentious now just wait.....
 
  • Like
Reactions: Curmudgeon666

Det0x

Golden Member
Sep 11, 2014
1,027
2,953
136
Do you realize PBO = Turbo? Turbo performance doesn't come free, you know?
Intel turboboost is not the same as AMD PBO... Instead of letting the cpu use ~2x the rated power ("TDP") in tau periode you can tell the cpu to boost maximum in a preset packet power target, and pbo will never go above it.

6 cores limited to 65w package power with PBO curve optimizer
6 core stock 65w -30 co.png

16 cores limted to 120w package power with PBO curve optimizer
16 core stock 120w -30 co.png

PBO normally controll 3 power limiters, edc tdc and ppt.
If you set PPT to a high/unlimited number the cpu will try to boost as high as it can with no regards to power (ppt), with the only limiting factors being temp and FIT value. But it can just as well be used like i showed above where you set a maximum packet power target and it will always stay below it.
 
Last edited:

Rigg

Senior member
May 6, 2020
435
863
106
Certainly not all power limits? What happened to the power cap? The 5800x enjoys the same power cap as the 5900x and 5950x? You've been bashing Intel for ages about releasing unrestricted power chips for ages, now all of a sudden AMD also does it?
FYI

There is a clear distinction between how AMD and Intel handle power limits. Intel makes a 'recommendation' and its AIB partners do whatever they want for default power settings in the BIOS. AMD has a standard and requires the AIB's to stick to it. PBO is a clearly defined overclocking feature that can only be user enabled, and comes with warnings when it is enabled. The problem isn't whether or not the CPU's have optional power limits it's how they are standardized. Intels approach on this front is not good for inexperienced end users. Escpecially when this is not clearly explained/disclosed by reviewers.
 

Abwx

Lifer
Apr 2, 2011
10,847
3,297
136
Certainly not all power limits? What happened to the power cap? The 5800x enjoys the same power cap as the 5900x and 5950x? You've been bashing Intel for ages about releasing unrestricted power chips for ages, now all of a sudden AMD also does it?

Same power limit for all SKUs mean that the 5950X doesnt exceed 4.05GHz in MT, if they had gone the Intel route they would had maxed out the 16C at close to 2x the 5800X power, something like 250W during 10-30s...
 
  • Like
Reactions: Tlh97

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
I think you're both beating about the bush. Both systems allow the end user to run the chips beyond stock power. Do you remember the 95w limited tests of the 9900k? Because forum warriors can't bear the thought of the Intel flagship at the time, running beyond spec. Intel's flagship was the 95w chip while AMDs chips were running at 125w at stock. Turbo is also time limited, while PBO boosts opportunistically. All these things have gone unnoticed because some motherboard manufacturers have determined they could run Intel chips at a certain power indefinitely. This is doesn't even take into consideration that Intel is the one who has strictly locked down non-K chips. Also, most K chip owners buy those chips to overclock them. Meanwhile, Ryzen owners are all over the place trying to squeeze every iota of performance from DRAM calculator and Curve Optimizer. The irony.
 

Makaveli

Diamond Member
Feb 8, 2002
4,715
1,049
136
Well, either the chip is throttling power or PBO is off or this is a heavily tweaked system (Curve Optimizer?).

Yes my system is not stock.

PBO+CO + Tuned memory. Goal here is performance I can drop it back to stock and XMP memory not trying to conserve power on my rig. If that was the idea I would just enable eco mode in the bios which will force it to 65watt TDP. As stated before I posted to show the difference in power consumption between Aida and cinebench. Since I have two extra cores to the 5600X not a direct comparison.
 
  • Like
Reactions: Zucker2k

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
Same power limit for all SKUs mean that the 5950X doesnt exceed 4.05GHz in MT, if they had gone the Intel route they would had maxed out the 16C at close to 2x the 5800X power, something like 250W during 10-30s...
And a supply of ln2 most likely? Ryzen benefits more from undervolting than what you're suggesting. I think you should know better. These chips run hot at stock power, much less double the power?
 

Hitman928

Diamond Member
Apr 15, 2012
5,177
7,628
136
I think you're both beating about the bush. Both systems allow the end user to run the chips beyond stock power. Do you remember the 95w limited tests of the 9900k? Because forum warriors can't bear the thought of the Intel flagship at the time, running beyond spec. Intel's flagship was the 95w chip while AMDs chips were running at 125w at stock. Turbo is also time limited, while PBO boosts opportunistically. All these things have gone unnoticed because some motherboard manufacturers have determined they could run Intel chips at a certain power indefinitely. This is doesn't even take into consideration that Intel is the one who has strictly locked down non-K chips. Also, most K chip owners buy those chips to overclock them. Meanwhile, Ryzen owners are all over the place trying to squeeze every iota of performance from DRAM calculator and Curve Optimizer. The irony.

I think you are getting confused between Precision Boost (PB) and Precision Boost Overdrive (PBO).
 

Abwx

Lifer
Apr 2, 2011
10,847
3,297
136
And a supply of ln2 most likely? Ryzen benefits more from undervolting than what you're suggesting. I think you should know better. These chips run hot at stock power, much less double the power?

Power is concentrated in a single chiplet for the 5800X, with two chiplets the 5950X run much cooler due to better power distribution.

That being said i m curious about ADL since the P cores area is more or less comparable to Zen 3, according to the recent leaks it did run at 108°C, such temp is not sustainable for long periods even if silicon limits are much higher (150-175°C).
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
Power is concentrated in a single chiplet for the 5800X, with two chiplets the 5950X run much cooler due to better power distribution.

That being said i m curious about ADL since the P cores area is more or less comparable to Zen 3, according to the recent leaks it did run at 108°C, such temp is not sustainable for long periods even if silicon limits are much higher (150-175°C).
According to the overclocker that 108c was on air, and cooler wasn't bolted down. I suppose we shall soon see.
 

Rigg

Senior member
May 6, 2020
435
863
106
I think you're both beating about the bush. Both systems allow the end user to run the chips beyond stock power.
Yes. But one is guaranteed to run at 'stock' power limits out of the box while the other isn't. This may be no big deal to an enthusiast but an inexperienced user who doesn't understand how power and turbo duration limits work on intel CPU's could be faced with a system that either severely under performs review benchmarks or generates much more heat and power usage than expected.

Do you remember the 95w limited tests of the 9900k? Because forum warriors can't bear the thought of the Intel flagship at the time, running beyond spec. Intel's flagship was the 95w chip while AMDs chips were running at 125w at stock. Turbo is also time limited, while PBO boosts opportunistically. All these things have gone unnoticed because some motherboard manufacturers have determined they could run Intel chips at a certain power indefinitely.
TDP is almost completely meaningless with modern CPU's. AMD and Intel don't even define TDP the same way and it is not comparable between brands. Intel's definition of TDP is much more sensible than AMD's but they rate the TDP at base clocks so calling a 9900k a 95w chip means almost nothing. That CPU is only guaranteed to be at or below 95w at base clock. The majority of Z chipset motherboards run at max all core turbo indefintely and ignore any realistic power limits right out of the box.

This is doesn't even take into consideration that Intel is the one who has strictly locked down non-K chips
Non K chips are not locked on power or turbo duration. Only their frequency is locked.

Also, most K chip owners buy those chips to overclock them.
Based on what? I'd wager the opposite is probably true but I don't think either one of has data to back up our assumptions.

Meanwhile, Ryzen owners are all over the place trying to squeeze every iota of performance from DRAM calculator and Curve Optimizer. The irony.

I fail to see the irony.
 

diediealldie

Member
May 9, 2020
77
68
61
CPU efficiency, either it's area or power, are not really a simple thing to measure. In fact, they're related. Sometimes, area efficiency helps you achieve power efficiency...etc.

In many cases, P cores are more power-efficient than E cores. That's how they're designed. For example, if you're in complicated branch conditions with huge memory IO then heavily stuffed cores will give you efficiency by reducing access to DRAM(less memory bus usage), branch misses(less decoder usage)...etc.

But there are some cases which allow E cores to be more efficient than P cores.

- Polling idle loops
Your thread is waiting for others to finish their jobs so practically processing instructions mean nothing here. It's possible to free Coves when the thread is in polling idle loops

- Throughput computing
You have a limited silicon budget(size). If you pack Gracemonts to the same die size + same power envelope then throughputs will be much greater compared to a pack of Coves which has unnecessary circulatories for this task(That's how GPUs are designed).

(TL;DR) P cores are 'really' fast and efficient when it comes to single-complicated tasks(heavy branch predictor and OoO window helps). When it comes to throughput workloads, E cores are 'really' efficient in both area and power(heavy branch predictor and OoO window are burdens).
 
Last edited:

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
Certainly not all power limits? What happened to the power cap? The 5800x enjoys the same power cap as the 5900x and 5950x? a

Dude

PBO = overclocking

Period.

In order to enable it, you have to click past "YOU MAY DAMAGE YOUR HARDWARE" warnings, both in the UEFI and in Ryzen Master. Pretty sure it can void your warranty on the CPU too.

You've been bashing Intel for ages

Can't let Charlie have all the fun.

btw thanks for the thread derail.