Anands Core-X power consumption results unique on the Internet

[Phynaz]

Purch Group is owner of Anandtech (also Tom's Hardware) and as far as I know they have rather close relationship with Intel.
https://en.wikipedia.org/wiki/Purch_Group
http://www.purch.com/experiences/intel/

Put away the tin foil hat. Intel spends hundreds of millions of dollars on advertising every year. Any ranked website runs their ads.

Ian's reply to the dip**** in the review:

Just think about what you're saying. If I was in Intel's pocket, we wouldn't be being sampled by AMD, period. If they were having major beef with how we were reporting, I'd either be blacklisted or consistently on a call every time there's been an AMD product launch (and there's been a fair few this year).

I've always let the results do the talking, and steered clear from hype generated by others online. We've gone in-depth into the how things are done the way they are, and the positives and negatives as to the methods of each action (rather than just ignoring the why). We've run the tests, and been honest about our results, and considered the market for the product being reviewed. My background is scientific, and the scientific method is applied rigorously and thoroughly on the product and the target market. If I see bullshit, I point it out and have done many times in the past.

I'm not exactly sure what you're problem is - you state that the review is 'slanted journalism', but fail to give examples. We've posted ALL of our review data that we have, and we have a benchmark database for anyone that ones to go through all the data at any time. That benchmark database is continually being updated with new CPUs and new tests. Feel free to draw your own conclusions if you don't agree with what is written.

Just note that a couple of weeks ago I was being called a shill for AMD. A couple of weeks before that, a shill for Intel. A couple before that... Nonetheless both companies still keep us on their sampling lists, on their PR lists, they ask us questions, they answer our questions. Editorial is a mile away from anything ad related and the people I deal with at both companies are not the ones dealing with our ad teams anyway. I wouldn't have it any other way.

 

[LTC8K6]

Is this much ado about nothing? Package power comes from processor sensors; variance from different methods of measurement.

You can view package power in HWiNFO (Windows).

Ah but they are clearly riding the conspiracy train, so the package power sensors are calibrated to lie...

And no one has noticed...

Hold on, let me see who owns Kill-a -watt, they might be in on it, too.

 

[eddman]

7900X consumes less:

anandtech
bit-tech
legitreviews
pcper
hardwarecanucks
hothardware
nordichardware
hardware.fr

About the same:

computerbase.de

1920X consumes less:

techspot
tweaktown

 

[Abwx]

AT's figures are in line with other reputable reviewers, such as Hardware.fr's:

Prime95 28.10 as the workload, similar to AT.

Hardware.fr figures include VRM losses, whereas AT's figures do not.

http://www.hardware.fr/articles/964-8/overclocking-pratique.html

Yes but those numbers are with a 256K FFT, according to the same Hardware.fr they get values 25-30% higher with Intel CPUs than what is displayed above if they use smaller FFTs, and in those cases TDP is higher by as much (than this delta) than the official figures.

 

[Dufus]

TDP is a rating. If you buy tires for your car rated at 100MPH and drive that car at 120MPH it doesn't change the rating to 120MPH.

And Intel power estimations are just that.and usually work fairly well if the CPU is operated within specification.

 

[tential]

7900X consumes less:

anandtech
bit-tech
legitreviews
pcper
hardwarecanucks
hothardware
nordichardware
hardware.fr

About the same:

computerbase.de

1920X consumes less:

techspot
tweaktown

OP started this topic, and so far, I'm being convinced of the EXACT opposite. I went in actually ahving the belief that TR consumed less power. Came out being more willing than ever to spend the intel premium,

 

[Arachnotronic]

OP started this topic, and so far, I'm being convinced of the EXACT opposite. I went in actually ahving the belief that TR consumed less power. Came out being more willing than ever to spend the intel premium,

Skylake-X is a darn nice platform. No regrets going with it across multiple systems.

 

[Bouowmx]

`TDP = electric power` is useful only when motherboard has power limit set to TDP. Most Intel X and Z motherboards have power limit set very high (> 1 kW); "140 W" is an idle suggestion then. User can set power limit back to 140 W, if desired.

ASUS ZENITH EXTREME has power limit 180 W at default; not sure if other X399 motherboards have power limit.

Power limits are fine, only when the setting is easily configurable. Read ASUS ZENITH EXTREME's manual for this setting..

 

[urvile]

Wow. I didn't know this forum is owned by an online marketing company. Seems like a smart strategy for purch actually. I wonder if they bought the forum to make money selling marketing to tech companies or if they are just really nice guys?

I should point out I don't actually care who owns the forum. It seems to me there is a lot of bias from tech reviewers generally. Going in all directions too.

Oh and I don't give a shit about intel tdp either. Actually I can think of a couple of reviewers on wccftech. One is a unashamed nvidia fanboy and the other is a foaming at the mouth amd fanboy.

Maybe that blances it out?

 

[Jan Olšan]

7900X consumes less:

anandtech
bit-tech
legitreviews
pcper
hardwarecanucks
hothardware
nordichardware
hardware.fr

About the same:

computerbase.de

1920X consumes less:

techspot
tweaktown

Two more I recall where 1950X ended up better in power consumption are GamersNexus (12V rail measurement, not wall) and Tom's HW (ditto) - their bad results with 7900X are probably well known already, their mobo/sample didn't hold TDP and they test in Prime95, so ouch.

Edit: I remembered third: SweClockers. They have wall numbers (so less conclusive) for Blender and x264 and while they trade places, on average 1950X is slightly better in load, despite haveing worse idle. http://www.sweclockers.com/test/24180-amd-ryzen-threadripper-1950x-och-1920x/22#content Besides these 3 I haven't read any other reviews that aren't on your list already, I think, so I have no idea if there might be more with results as these.

Further edit: And the delayed Tech Report review
Could be put in the "about the same" category, but 1950X did end up consuming * 1W * less at wall in Blender.

Naturally this might change if other boards not used in these reviews don't enforce TDP on Threadripper, though.

 

[BigDaveX]

I'd expect there to be a lot more variance in LGA2066 results simply because there's a whole bunch of different motherboards out there, versus only about a half-dozen or so Socket TR4 boards.

 

[Yuriman]

I'd expect there to be a lot more variance in LGA2066 results simply because there's a whole bunch of different motherboards out there, versus only about a half-dozen or so Socket TR4 boards.

Wouldn't surprise me.

A few years back, I had a couple of Z77 boards kicking around, and the power consumption at the wall with my i5, at the same clocks, was 40w different idle and load between the highest and lowest consuming boards. This is a 77w TDP CPU, mind you.

 

[eddman]

OP started this topic, and so far, I'm being convinced of the EXACT opposite. I went in actually ahving the belief that TR consumed less power. Came out being more willing than ever to spend the intel premium,

Intel i9 processors seem to be quite good in performance and perf/watt aspects, but IMO they are simply way too expensive for what they offer. I mean, pricing a 10 core part same as AMD's 16 core? The numbers of PCI lanes could also be a factor for certain buyers.

I think the following would be more fair towards buyers:

7900X: $800
7920X: $900
7940X: $1000
7960X: $1200 to $1300
7980XE: $1400 to $1600

I would certainly like to have one, but just can't wrap my head around their pricing.

 

[Arachnotronic]

Intel i9 processors seem to be quite good in performance and perf/watt aspects, but IMO they are simply way too expensive for what they offer. I mean, pricing a 10 core part same as AMD's 16 core? The numbers of PCI lanes could also be a factor for certain buyers.

I think the following would be more fair towards buyers:

7900X $800
7920X $900
7940X $1000
7960X $1200 or $1300
7980XE $1400 or $1600

The 10 core part has faster cores with much more OC potential than TR does; it's a more balanced system.

And honestly, 60 PCIe lanes vs 44? Total overkill for most consumers. I'm nowhere near saturating the 44 lanes on my 7900X, and even my 28 lane systems still have some free CPU PCIe lanes left.

 

[raghu78]

The 10 core part has faster cores with much more OC potential than TR does; it's a more balanced system.

And honestly, 60 PCIe lanes vs 44? Total overkill for most consumers. I'm nowhere near saturating the 44 lanes on my 7900X, and even my 28 lane systems still have some free CPU PCIe lanes left.

For the kind of audience and workloads that TR 1950x and 7900x are targetted the higher single thread performance of 7900x really does not mean much. For OC headroom a 7900x at stock runs at 4 Ghz all core turbo while 1950x runs at 3.5 Ghz all core turbo. 7900X can do 4.6 Ghz while 1950X can do 4 Ghz on avg. So how exactly does 7900x have much more OC headroom. Delidding a 7900x might get you 200 Mhz more. But voiding the warranty is seriously not an option for most users on such expensive hardware.

For megatasking and multitasking a 1950x kills a 7900x. The perf gap widens with mega tasking and multi tasking.

https://www.youtube.com/watch?v=EsqqCvhadWM

7900X cannot even be considered a faster CPU in games against 1950X as Skylake-X game performance is as inconsistent as Ryzen TR.

http://www.hardware.fr/articles/967-18/indices-performance.html
https://www.youtube.com/watch?v=zd7Sn80Og40

We have already seen 7800x at 4.6 Ghz pretty much tied with a 1600 at 4 Ghz for gaming. Skylake-X is not an all round better gaming CPU than Ryzen.

https://www.techspot.com/review/1450-core-i7-vs-ryzen-5-hexa-core/page9.html

So the 7900x is not a more balanced system than 1950x. Skylake-X makes tradeoffs similar to Ryzen which benefit multithread performance at the cost of latency and game performance.

 

[MarkPost]

Power consumption isnt an issue per se, HEAT IS the issue. And the whole Skylake-X line suffers a ton of that (cant imagine how this will be when 12, 14, 16 and 18 models are released). TR hasnt this issue (not specially), even being testing yet with no specific TR coolers. So it only became even better once proper coolers are available.

 

[ehume]

I can understand that the operators of servers and server farms care about current draw and stuff like that -- it means money to them. As a once and future reviewer of cooling solutions I don't care what Kill-A-Watt says. I care about how much heat I need to take away. For that I am not concerned about what the system draws but what the CPU draws.

Really, the more current the CPU draw the happier I am. I was happy to see that the Skylake series has a high TDP. I was disappointed to see the TDP numbers attached to the Coffee Lake series. To really test a heatsink I would a CPU with an iGPU, not needing a dGPU. It seems that dGPU fans make noise. Even passive ones make heat.

 

[lobz]

Efficiency goes out of the toilet once you put moar cores into Ryzen. Even worse when you up the clockspeed.

efficiency? for that 30W more power the 1950x beats the hell out of the 7900x in cinebench

 

[Gikaseixas]

Both are beasts and as such, power consumption while important, will take a back seat to other aspects. Let's not fight as if this is the end of the world guys. Intel is deservedly getting some heat due to the tin scandal and that's that.

Threadripper is the better chip at the moment IMO, when you consider every performance, thermal management parameters.

 

[urvile]

The 10 core part has faster cores with much more OC potential than TR does; it's a more balanced system.

And honestly, 60 PCIe lanes vs 44? Total overkill for most consumers. I'm nowhere near saturating the 44 lanes on my 7900X, and even my 28 lane systems still have some free CPU PCIe lanes left.

The main market for HEDT isnt enthusiasts it's professional users in enterprise and smaller businesses. They don't give a rats arse about over clocking. It doesnt even factor in to their considerations. Neither do things like number of PCIe lanes.

They just want hardware that meets their requirements at a reasonable price. In enterprise at least in my experience the upgrade process just involves replacing the current workstation with a newer more powerful one every 12 to 18 months.

 

[LTC8K6]

efficiency? for that 30W more power the 1950x beats the hell out of the 7900x in cinebench

Server farms don't run Cinebench, though, and we would expect 6 more cores to dominate in a multi-thread benchmark.

You also have to balance power and heat with the amount of work done. If a chip gets more work done in a day, that can offset any power and heat differences. It can also remove any price questions.

 

[lobz]

Server farms don't run Cinebench, though, and we would expect 6 more cores to dominate in a multi-thread benchmark.

You also have to balance power and heat with the amount of work done. If a chip gets more work done in a day, that can offset any power and heat differences. It can also remove any price questions.

Server farms?? In all honesty I thought this was about Ryzen Threadripper versus Sylake-X i7 7900 desktop processors.
Tlking about epyc vs the new metallic xeons, sometimes you will find the most interesting results around the web.
Just an example: https://www.heise.de/ct/ausgabe/201...6148-E5-2699v4-und-AMD-Epyc-7601-3787494.html

most important findings are:

-2X Epyc 7601 and 2X Xeon Platinum 8180 are extremely close. Xeon was 12% ahead in int_rate_base and 4.5% ahead in fp_rate_base.

-Without NUMA optimiztion, Epyc is ahead in STREAM-Triad for very low thread counts(<8) sometimes by almost 2X over Skylake-SP Xeons. The 8180 is bandwidth-starved compared to the 6148 by 35% for 16-32 threads. When fully saturated with the maximum permissible number of threads, bandwidth on Skylake-SP dips while Epyc remains constant.

-The huge performance disparity in libquantum is obliterated when compiled for 64-bit. Compile times were reduced to two seconds from 10 minutes on 32-bit mode. Yes, minutes. Epyc and Xeon 8180 give the same performance even with AVX512 enabled on the latter in 64-bit mode.

 

[LTC8K6]

Server farms?? In all honesty I thought this was about Ryzen Threadripper versus Sylake-X i7 7900 desktop processors.
Tlking about epyc vs the new metallic xeons, sometimes you will find the most interesting results around the web.
Just an example: https://www.heise.de/ct/ausgabe/201...6148-E5-2699v4-und-AMD-Epyc-7601-3787494.html

most important findings are:

-2X Epyc 7601 and 2X Xeon Platinum 8180 are extremely close. Xeon was 12% ahead in int_rate_base and 4.5% ahead in fp_rate_base.

-Without NUMA optimiztion, Epyc is ahead in STREAM-Triad for very low thread counts(<8) sometimes by almost 2X over Skylake-SP Xeons. The 8180 is bandwidth-starved compared to the 6148 by 35% for 16-32 threads. When fully saturated with the maximum permissible number of threads, bandwidth on Skylake-SP dips while Epyc remains constant.

-The huge performance disparity in libquantum is obliterated when compiled for 64-bit. Compile times were reduced to two seconds from 10 minutes on 32-bit mode. Yes, minutes. Epyc and Xeon 8180 give the same performance even with AVX512 enabled on the latter in 64-bit mode.

I'm sure we can all post benches that support whatever we want to support.
I'm sure the mfgs can show us benches that support whatever they want us to believe.

And you don't need to try to sell me anything.

At any rate, I'm not buying anything but regular desktop chips for the foreseeable future.

 

[swilli89]

Purch Group is owner of Anandtech (also Tom's Hardware) and as far as I know they have rather close relationship with Intel.
https://en.wikipedia.org/wiki/Purch_Group
http://www.purch.com/experiences/intel/

This should answer many questions some have.

 

[jmelgaard]

Obviously...

7900X consumes less:

anandtech - fake news
bit-tech - fake news
legitreviews - fake news
pcper - fake news
hardwarecanucks - fake news
hothardware - fake news
nordichardware - fake news
hardware.fr - fake news

About the same:

computerbase.de - fake news

1920X consumes less:

techspot - totally credible, best news
tweaktown - totally credible, best news

We all know that!

(Sorry.... could not help it >.<....)

Anyways, it would be much more constructive to look into the differences and put a light on how it is measured, what was the system configuration, what loads was the CPU under etc... Comparing results across sites with different testing methologies for the same thing just doesn't make any sense...