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.
There's that, too, but I like to highlight the process because that is the foundation for everything else. The two are intrinsically linked.This isn't a process thing, but rather an architecture one.
Not at all. The process can be worse while certain architectural choices (e.g. no separate IO die, higher IPC core) make the overall product better.There's that, too, but I like to highlight the process because that is the foundation for everything else. The two are intrinsically linked.
This isn't a process thing, but rather an architecture one.
I am not going to go into every test, and try to prove my point, but just let me say that those tests were flawed in many ways, and biased. For one, the 5950x with PBO set to auto ? (which it must have used at over 200 watts on a couple of tests) And what appeared to be a lot of single core tests. I grant that AL is better single core, but not sure at what power level.So, I know these observations were hard to believe at first, and rightly so, based on recent history, but Intel 7 is actually more efficient than TSMC 7N based on the final apple to apple (aka core to core) comparison done by Igor's Lab. The i5-12400 outperforms the R5-5600x in both games and productivity tests while using significantly less power.
It is pretty simple, i5 12400 will eat more power at 4.5ghz all core turbo.
- R5 5600X, 4.5ghz all core turbo
- i5 12400, 4ghz all core turbo
So, I know these observations were hard to believe at first, and rightly so, based on recent history, but Intel 7 is actually more efficient than TSMC 7N based on the final apple to apple (aka core to core) comparison done by Igor's Lab. The i5-12400 outperforms the R5-5600x in both games and productivity tests while using significantly less power.
As I said above, he set the motherboard PBO to auto, so you know its engaging. He is biased IMO, by setting things to make the 12400 look better. I DO agree that if you set the power limits reasonable, alder lake is a decent CPU, but especially the 12900k and less so the others, its set too high by default. So he sets everything low for AL, and high on AMD.That s flawed datas since he "measures" the 5600X at 90W in Blender but Computerbase measure about 70W in Cinebench wich is more power consuming, even with Prime 95 they dont get higher than 76W CPU package power.
Intel Core i9-12900K, i7-12700K & i5-12600K im Test: Leistungsaufnahme, Temperaturen und OC
Intel Alder Lake im Test: Leistungsaufnahme, Temperaturen und OC / Leistungsaufnahme in Spielen / Leistungsaufnahme in Anwendungenwww.computerbase.de
As I said above, he set the motherboard PBO to auto, so you know its engaging. He is biased IMO, by setting things to make the 12400 look better. I DO agree that if you set the power limits reasonable, alder lake is a decent CPU, but especially the 12900k and less so the others, its set too high by default. So he sets everything low for AL, and high on AMD.
As you said, flawed data.
To repeat (for those who think I am biased) When Alder lake has reasonable power limits set, its a decent CPU. I almost got one, but for my use case the 5950x is better. The limits for the 12900k are just insane IMO, and setting the motherboard to "auto on PBO" for AMD is almost as bad. PBO should NEVER be enabled for Ryzen and PL1 and PL2 should be set to similar type reasonable power limits. I would really like for someone to do this in their testing, and do a full set of benchmarks of both in this. I have not seen this yet.Thanks, i wasnt aware of this trickery, most funny is that in the TW post above the 12400F is said to use 90W in CB R23...
Talk of inverting the truth..
So, the iso power characteristics of the process doesn't matter? I'm tempted to mention Cypress Cove but I have a feeling I'm going to be wrong again.Not at all. The process can be worse while certain architectural choices (e.g. no separate IO die, higher IPC core) make the overall product better.
The 12400 should embarrass the 5600x at same clocksIt is pretty simple, i5 12400 will eat more power at 4.5ghz all core turbo.
- R5 5600X, 4.5ghz all core turbo
- i5 12400, 4ghz all core turbo
Are you arguing about the 5950x efficiency in what is a head to head between equally (core/smt) matched 12400 and 5600x? I didn't even pay attention to any other chip besides those two, to be frank.I am not going to go into every test, and try to prove my point, but just let me say that those tests were flawed in many ways, and biased. For one, the 5950x with PBO set to auto ? (which it must have used at over 200 watts on a couple of tests) And what appeared to be a lot of single core tests. I grant that AL is better single core, but not sure at what power level.
This is not fact. I've seen a lot of reviews use Blender for power consumption tests because it consumes more power than both CB and P95 on Ryzen.That s flawed datas since he "measures" the 5600X at 90W in Blender but Computerbase measure about 70W in Cinebench wich is more power consuming, even with Prime 95 they dont get higher than 76W CPU package power.
Intel Core i9-12900K, i7-12700K & i5-12600K im Test: Leistungsaufnahme, Temperaturen und OC
Intel Alder Lake im Test: Leistungsaufnahme, Temperaturen und OC / Leistungsaufnahme in Spielen / Leistungsaufnahme in Anwendungenwww.computerbase.de
It matters, but one chip being more efficient than another chip does not necessarily mean that the former is on a better process.So, the iso power characteristics of the process doesn't matter? I'm tempted to mention Cypress Cove but I have a feeling I'm going to be wrong again.
What I am saying is that since BOTH AMD and Alder lake in these tests have the WRONG power usage, and the AMD one has wrong bios settings(maybe the alder lake also), all data, power usage and benchmark results are flawed and unusable IMO.Are you arguing about the 5950x efficiency in what is a head to head between equally (core/smt) matched 12400 and 5600x? I didn't even pay attention to any other chip besides those two, to be frank.
So, the iso power characteristics of the process doesn't matter? I'm tempted to mention Cypress Cove but I have a feeling I'm going to be wrong again.
The 12400 should embarrass the 5600x at same clocks
Are you arguing about the 5950x efficiency in what is a head to head between equally (core/smt) matched 12400 and 5600x? I didn't even pay attention to any other chip besides those two, to be frank.
This is not fact. I've seen a lot of reviews use Blender for power consumption tests because it consumes more power than both CB and P95 on Ryzen.
Gamer's Nexus measurements in blender show lower power use for the 5600x at stock as well. The 5600x really shouldn't be using 90W+ at stock.
Regardless of that.... Setting PBO to automatic, and using more power on the 5950x than any other site ? Right there they screwed up, and it affects all the AMD numbers and benchmark results. Also, I doubt the ADL power numbers as well for the same reason.Gamer's Nexus power test equipment: HWiNFO64
Igorslab power test equipment: special laboratory
Oscilloscope-based system:
Non-contact direct current measurement on PCIe slot (riser card)
Non-contact direct current measurement at the external PCIe power supply
Direct voltage measurement at the respective connectors and at the power supply unit
2x Rohde & Schwarz HMO 3054, 500 MHz multichannel oscilloscope with memory function
4x Rohde & Schwarz HZO50, current clamp adapter (1 mA to 30 A, 100 KHz, DC)
4x Rohde & Schwarz HZ355, probe (10:1, 500 MHz)
1x Rohde & Schwarz HMC 8012, HiRes digital multimeter with memory function
Do you see the difference? Unless both tests do exactly the same it's hard to compare.
Gamer's Nexus power test equipment: HWiNFO64
Igorslab power test equipment: special laboratory
Oscilloscope-based system:
Non-contact direct current measurement on PCIe slot (riser card)
Non-contact direct current measurement at the external PCIe power supply
Direct voltage measurement at the respective connectors and at the power supply unit
2x Rohde & Schwarz HMO 3054, 500 MHz multichannel oscilloscope with memory function
4x Rohde & Schwarz HZO50, current clamp adapter (1 mA to 30 A, 100 KHz, DC)
4x Rohde & Schwarz HZ355, probe (10:1, 500 MHz)
1x Rohde & Schwarz HMC 8012, HiRes digital multimeter with memory function
Do you see the difference? Unless both tests do exactly the same it's hard to compare.
So, the iso power characteristics of the process doesn't matter? I'm tempted to mention Cypress Cove but I have a feeling I'm going to be wrong again.
The 12400 should embarrass the 5600x at same clocks
Are you arguing about the 5950x efficiency in what is a head to head between equally (core/smt) matched 12400 and 5600x? I didn't even pay attention to any other chip besides those two, to be frank.
This is not fact. I've seen a lot of reviews use Blender for power consumption tests because it consumes more power than both CB and P95 on Ryzen.
Being able to read is often an immense benefit...If a 5600X on it's own is pulling 90W it is NOT stock. End of conversation.
It has a hard limit of a 76W PPT. You cannot get the 5600X to surpass that under stock operation.
Watts is a current draw measure. For total energy used, you have to use Joules. Understanding science is even better than just reading.Being able to read is often an immense benefit...
it's the power the CPU used FOR THE WHOLE DURATION OF THE RENDER (over complete render)
It's how much power it used to make all the graphics go ooh.
It's 88 W PPT. The formula is 1.35*TDP.If a 5600X on it's own is pulling 90W it is NOT stock. End of conversation.
It has a hard limit of a 76W PPT. You cannot get the 5600X to surpass that under stock operation.
Just to be clear, I'm not saying the 12400 isn't a great CPU, it is. It looks like it is basically a tie with a 5600x in multi-threaded performance and gaming while being a little faster in single core productivity. It may use slightly more power fully loaded (according to Tom's when 5600x PBO is off) but not that much more. The big win is on price and it makes the 5600x look overpriced. It just needs more budget motherboard options to make it really appealing.
The 5600X ships with a lower PPT for some weird reason. Not sure why.It's 88 W PPT. The formula is 1.35*TDP.
The AMD 3rd Gen Ryzen Deep Dive Review: 3700X and 3900X Raising The Bar
www.anandtech.com