Ryzen 3000: userbenchmark changed the score weights in favor of intel

[DrMrLordX]

Dunno what they will do with Zen 3, bringing MC to -20%..?..

Sounds about right.

 

[Space Tyrant]

Here's a nice rebuttal to the Userbenchmark "multicore beyond quadcore = 2% of gaming performance" position:
Note the 25% improvement gained going from 4 to 6 cores and the 7% improvement from 6 to 8 cores. I'm no mathematician, but these numbers seem to add up to considerably more than 2%.

I don't know how common this gap is in modern games -- or how how common 720p usage is -- but it would have to be quite rare to mathematically justify the 2% number.

 

[mopardude87]

Yes, the 3600X now comes with the Wraith Spire "2", a solid aluminum version of the original Spire -- which had a copper vapor chamber in the middle. The new spire was made to cool as effectively as the old one by turning up the fan speed by about 50%. So, yeah, banshee levels.

As far as I know, the other AMD coolers (stealth, prism) remain unchanged.

Featuring the new and improved stock cooler, with twice the cheapness and double the noise! Not liking the fact they changed the cooler AFTER its been out a while. A uninformed buyer will assume from original reviews that the chip will run fairly cool and quiet then you open the box and there's this dank cooler. You essentially are paying the same price for less now. It would be like Nvidia doing away with the double fan reference cooler,not announcing it and when you get it its a dang blower. Adding more noise to a build is more then a slight issue for some people.I can deal with some noise but the 8350 cooler makes more noise then i have ever been comfortable with out of a reference cooler.

Is there even a option to opt of a cooler to save a couple bucks or is everyone forced into paying a small premium for a paperweight? Intel is dumping lots of e-waste into the world offering their trash heatsinks with a i7 chip when everyone and their mother knows their heatsinks are as useless as gallon of gas in a house fire.

 

[Space Tyrant]

Is there even a option to opt of a cooler to save a couple bucks or is everyone forced into paying a small premium for a paperweight?

Until some etailer starts selling bare cpus out of trays, no. The retail box may be the only way to buy a single unit for a long time.

BTW, the retail box doesn't mention the word "spire"; it refers to the "premium cooler" inside. Compared to the Intel box coolers, I suppose, it really is "premium" -- just not compared to an R5 1600.

AMD did make a distinction in a slide at one of the pre-release conventions where the 3600X was said to come with the "Wraith Spire 2".

 

[coercitiv]

Hi.

I've been asked to deliver a word of warning to the Anandtech forums. It reads as follows:

Keep this up and we'll introduce a CPUID ratio.

This post was not written under duress. Long live the quads!

 

[John Carmack]

What I hate is that trashy web sites like Userb**chmark and GPU**ss always end up at the top of search results and things like the subject of this thread are what make them so hazardous to less informed people.

 

[Ranulf]

Oh, it gets worse for these guys. Look at this lovely exchange over at this Hardware Unboxed video's YT comments:

Let's go there, shall we?

https://web.archive.org/web/2017011.../Faq/What-is-the-effective-CPU-speed-index/55

Date: January 19, 2017. Right before Ryzen 1xxx launched (March 2017).

Single Core: 30%
Quad Core: 50%
Multi Core: 20%

Looking at the next available snapshot for that site, which is June 6, 2017.... Multicore is down to 10%.

Single Core: 30%
Quad Core: 60%
Multi Core: 10%


So, Ryzen 1xxx prompted them to bring MC to 10%. Ryzen 3xxx, down to 2%.

As you can see their justification for the weights at both times was 3D gaming... when games have been routinely using and benefiting from more than 4C/8T lately (considering the 2% change was done just a few moments ago on this time scale). That argument just falls apart and the timing on both adjustments is more than suspicious.

Yeah. These people are the epitome of integrity. Integrity meaning bribery.

I saw this whole deal last night on HWUB's break down of it, all 40 minutes heh. That YT comment is the nail in the coffin for userbenchmark in my opinion. Not that I had really used them beforehand. I barely remembered they existed until this debate.

Also to further rebuke DomSeraph's earlier post, "I am using my i7 4790 over my dual X5670 workstations because four fast cores beats 12 slower ones for general use."

You're using a 4c8t processor not a 4 core one, like what most i5 chips are or were until the 8600's. Now they're 6c/6t. If we've learned anything from the Sandy Bridge era onward, its that hyper threading/SMT helps smooth things out and is the better long term investment. This was obvious by haswell/skylake and really obvious by ryzen gen 1 in 2017. The battles over cheap i5's vs ryzen 5 in the last two years was really about high ram costs for 3200 speed and stability. Intel technically won but well, these days the numbers don't lie that the r5 1600 does better in most circumstances over the 4c/4t i5 7600.

When it comes to the old dual core vs quad core arguments, it was more justifiable to go dual core until 2011 or so on price. Long term quad was better but you could argue waiting 1-3 more years to upgrade to that. My Q8400 (2.6ghz) wasn't really much faster as it were than my C2D e6600 (2.4ghz) and by the time a quadcore was noticably better, Sandy/IvyBridge+ was on the scene.

 

[Schmide]

I just thought of a unique corollary:

This metric could be more about Ice Lake than current gen processors.

Step one establish a baseline maximizing the value of the i9 while hiding a 4 core easter egg.

Step two release a 4 core processor a couple months down the road that exploits the inequality. i7 Ice Lake meets or beats i9









Step three: underpants

 

[darkswordsman17]

Oh, it gets worse for these guys. Look at this lovely exchange over at this Hardware Unboxed video's YT comments:

Let's go there, shall we?

https://web.archive.org/web/2017011.../Faq/What-is-the-effective-CPU-speed-index/55

Date: January 19, 2017. Right before Ryzen 1xxx launched (March 2017).

Single Core: 30%
Quad Core: 50%
Multi Core: 20%

Looking at the next available snapshot for that site, which is June 6, 2017.... Multicore is down to 10%.

Single Core: 30%
Quad Core: 60%
Multi Core: 10%


So, Ryzen 1xxx prompted them to bring MC to 10%. Ryzen 3xxx, down to 2%.

As you can see their justification for the weights at both times was 3D gaming... when games have been routinely using and benefiting from more than 4C/8T lately (considering the 2% change was done just a few moments ago on this time scale). That argument just falls apart and the timing on both adjustments is more than suspicious.

Yeah. These people are the epitome of integrity. Integrity meaning bribery.

Not only that, but their justification for such with the little analogy at the end there "One fox beats one hundred chickens." What? That analogy makes no sense even with regards to just speaking about foxes and chickens. Sure a fox can eat a chicken, but that's because it'd sneak up on them or have a group of them in a cage to go after. Its not because the fox is so much quicker/faster than the chickens. Its not like foxes or chickens are used for work of any sort (unless you're David the Gnome). It makes such little sense for comparing CPUs that it'd almost have to be coded language or something. Its one of the dumbest things I've ever seen in any talk/writing about microprocessors in my entire life.

 

[DrMrLordX]

Step three: underpants

Step four: ????
Step five: PROFIT

 

[Furious_Styles]

What I hate is that trashy web sites like Userb**chmark and GPU**ss always end up at the top of search results and things like the subject of this thread are what make them so hazardous to less informed people.

Agreed. They must be paying google to be the top results because they are garbage.

 

[maddie]

Not only that, but their justification for such with the little analogy at the end there "One fox beats one hundred chickens." What? That analogy makes no sense even with regards to just speaking about foxes and chickens. Sure a fox can eat a chicken, but that's because it'd sneak up on them or have a group of them in a cage to go after. Its not because the fox is so much quicker/faster than the chickens. Its not like foxes or chickens are used for work of any sort (unless you're David the Gnome). It makes such little sense for comparing CPUs that it'd almost have to be coded language or something. Its one of the dumbest things I've ever seen in any talk/writing about microprocessors in my entire life.

I took it to mean, "one fast core is better than 100 slow cores". I also agree with you that the language is idiotic.

I guess AMD could respond with "100 fighting roosters demolish one fast fox".

 

[DrMrLordX]

I guess AMD could respond with "100 fighting roosters demolish one fast fox".

I will take 12 velociraptors. Or perhaps terror birds.

 

[Kenmitch]

I will take 12 velociraptors. Or perhaps terror birds.

T-Rex always saves the day....At least according to Hollywood.

 

[maddie]

I will take 12 velociraptors. Or perhaps terror birds.

Hey, you work with what you have, or do you have some hidden island somewhere?

 

[Zucker2k]

T-Rex always saves the day....At least according to Hollywood.

Do I smell the codename for a threadripper successor?

 

[DrMrLordX]

Hey, you work with what you have, or do you have some hidden island somewhere?

The island is called: "Taiwan". I hear they fab semiconductor products there.

 

[moinmoin]

After Userbenchmark PassMark is the next benchmarking site making all its numbers useless by boosting some obscure parameters in favor of Intel chips:

PassMark CPU Benchmarks - Single Thread Performance

Benchmarks of the single thread performance of CPUs. This chart comparing CPUs single thread performance is made using thousands of PerformanceTest benchmark results and is updated daily.

www.cpubenchmark.net

The first AMD chip in the single thread performance listing is now AMD Ryzen 3 4300U at position 35...

Yesterday there were still some few more AMD chips: http://web.archive.org/web/20200312201105/https://www.cpubenchmark.net/singleThread.html
Until a week ago AMD chips dominated the top:
http://web.archive.org/web/20200309064753/https://www.cpubenchmark.net/singleThread.html

That's quite the strong push for some alternate reality.

 

[Markfw]

After Userbenchmark PassMark is the next benchmarking site making all its numbers useless by boosting some obscure parameters in favor of Intel chips:

PassMark CPU Benchmarks - Single Thread Performance

Benchmarks of the single thread performance of CPUs. This chart comparing CPUs single thread performance is made using thousands of PerformanceTest benchmark results and is updated daily.

www.cpubenchmark.net

The first AMD chip in the single thread performance listing is now AMD Ryzen 3 4300U at position 35...

Yesterday there were still some few more AMD chips: http://web.archive.org/web/20200312201105/https://www.cpubenchmark.net/singleThread.html
Until a week ago AMD chips dominated the top:
http://web.archive.org/web/20200309064753/https://www.cpubenchmark.net/singleThread.html

That's quite the strong push for some alternate reality.

Thats quite interesting that 2 sites now have a favorite for Intel. Suppose some money from Intel dissuaded them ?

 

[moinmoin]

It gets better, according to some Twitter response there was a change in how the testing is done on or before March 6th. But as the above web.archive.org links show the number still have been heavily tweaked after that point.

https://twitter.com/x/status/1235774287572799488

Edit: Going by the dates in the changelog either build 1002 or 1003 did these changes: https://www.passmark.com/products/performancetest/history.php
Only hint in the notes that may fit is
"CPU tests, Single Threaded, started scaling single threaded score down to be closer to PT9 for better comparability with older results."
Yeah, makes sense... not.

 

[DrMrLordX]

Lame