Intel wants reviewers to benchmark using windows media player instead of cinebench for low end mobile

[5x5]

I'm actually surprised intel didn't advocate 720p gaming benchmarks cause "that's what cs players want"

[orbitalbuzzsaw]

Fairness, North Korea style!

[CTR640]

You know what, I love the internet because Intel can no longer play dirty like they did in the Pentium4 era where the internet was not common. So they could manipulate the benchmarks and people believed that when they were in the stores. It's very good we got Der8auer, LTT and many more Youtuber journalists.

 

And to the subject of the topic, the first thing when I read the article: "Intel, are you trying to manipulate again?" This won't work. When I have to upgrade my CPU, it will be definitely AMD. Seriously, fuck Intel and they shady tactics.

 

And also Chrome??? LMFAO what the actual fuck??? That shit makes the map of GTAV disappear when I open Chrome. Firefox FTW!

[Thaldor]

2 hours ago, mr moose said:

The point is that sometimes those conclusions are drawn using workloads that the end user won't so the end conclusion if not properly qualified will be pointless. 

 

Some reviews are more than 50% production/tool workload based and if the conclusions and recommendations don't properly explain that (which many don't) then the only way to work out how good the CPU is for your specific task is to go through the whole review and work out which bits are irrelevant. 

It's just that you cannot throw "normal" use workloads at modern desktop CPUs and expect to get any kind of trustful or useful results without going synthetic. Even the mid-tier i5s and Ryzen 5s eat Chrome and Office for breakfast and the top-tier laughs at them and at that point it's just completely waste of time to test anything like that because the reason why Chrome opened that 0.0001 seconds slower with CPU Y compared to CPU X is more likely to be caused by Windows having a hiccup or even a bit being in a different place in the RAM than did you spend more on the CPU. Not to even talk about testing some most played games, currently I would remember CSGO runs with i9-9900K+RTX 2080 Ti in some +600FPS ranges when it's not restricted, at that point even the HDMI/DVI/DP-link connection starts to affect the results, not to even talk about how long distance there's between the CPU and GPU on the motherboard, that 3mm difference might be 2-3FPS less because latency (not to even mention some games that have shown that i7-7700K is still the king of the CPUs because for some magical reason they attain higher FPS with it (and that is called optimization)).

[Blademaster91]

1 hour ago, leadeater said:

I like to get from stand still to the speed limit rather quickly all the time . Not in a dangerous way but the woosh is fun sooo.... WOOOOOOOoooooooooooo.

It's fun especially if the car is geared well enough to break traction from a stop, not something I do anyway cuz its extra wear and tear. But off the line acceleration is something most people don't care about and isn't much fun in most cars with autos or crap CVT's.

[leadeater]

21 minutes ago, Blademaster91 said:

It's fun especially if the car is geared well enough to break traction from a stop, not something I do anyway cuz its extra wear and tear. But off the line acceleration is something most people don't care about and isn't much fun in most cars with autos or crap CVT's.

I leave my TC and ESC on always, not looking to buy a new car . But with that all on 0-60 in 5.3. Not looking to tear through BS RE050A's quickly either, too expensive.

[Curufinwe_wins]

In fairness to Intel, if someone could figure out an even remotely consistent office suite test that was more than opening the programs, I and a lot of other productivity focused individuals would love to see those done and it would actually be an influence in our buying decisions (because you haven't been living if your excel isnt on the verge of crashing every day at work due to being stuck with shitty hardware).

 

And yes, solution is partially "don't use excel", but python matlab (both of which I also use) whatever live filtering is much faster with pivot tables, even if less robust.

[porina]

13 hours ago, bcredeur97 said:

Is it just me or is cinebench actually one of the most fair tests out there?? You’re just rendering a picture...

Different types of code will behave differently, and be influenced by the rest of the system differently. Cinebench R15 has a characteristic of scaling near perfectly with cores, has a high benefit from HT/SMT, and is minimally affected by ram performance. Haven't got to know R20 as well yet, but it seems pretty similar. It is near enough the ideal case to show off peak CPU performance. 

 

9 hours ago, imreloadin said:

Have people seriously forgotten that Cinebench is the benchmark for the Maxon Cinema software? Like it's actually a benchmark for a real world 3d rendering software...

That how many people actually use? If you're in that niche, great. If you're not, how much value does it really provide you?

 

2 hours ago, Taf the Ghost said:

They also miss that R20 shifted to a lot more AVX2 workload, which improve Intel vs AMD in Intel's favor. At least until Zen2 launched. 

That doesn't really change the number of people who actually use the product though.

 

 

I run Cinebench as much as the next person on the forums. As said, it is useful as an indication of peak scaling potential, but it certainly wont be representative of software that is more demanding of resources. I'd like to see it kept in reviews, but its importance shouldn't be over-stated.

[nick name]

17 hours ago, For Science! said:

I'm more concerned about the 0.22% running cinema4D on their tablets.....

I'm pretty sure they only do it for science.  

[Bcat00]

5 hours ago, RejZoR said:

But really, do you really need a 9900K to run a... BROWSER?

Nope, definitely nope

[Taf the Ghost]

25 minutes ago, porina said:

Different types of code will behave differently, and be influenced by the rest of the system differently. Cinebench R15 has a characteristic of scaling near perfectly with cores, has a high benefit from HT/SMT, and is minimally affected by ram performance. Haven't got to know R20 as well yet, but it seems pretty similar. It is near enough the ideal case to show off peak CPU performance. 

 

That how many people actually use? If you're in that niche, great. If you're not, how much value does it really provide you?

 

That doesn't really change the number of people who actually use the product though.

 

 

I run Cinebench as much as the next person on the forums. As said, it is useful as an indication of peak scaling potential, but it certainly wont be representative of software that is more demanding of resources. I'd like to see it kept in reviews, but its importance shouldn't be over-stated.

The data was from 15w laptops and tablet form factors. Once 7nm Zen2 APUs are out, I expect all of those programs to be in AMD's favor.

 

Cinebench is relative to a huge chunk of professional workloads, so it works very well, as a benchmark, for testing. 

[porina]

18 minutes ago, Taf the Ghost said:

Cinebench is relative to a huge chunk of professional workloads, so it works very well, as a benchmark, for testing. 

It isn't an area I've looked into closely, but what would be a "professional workload"? I don't recall any reputable reviewer saying words to the effect "It runs Cinebench great, so it'll also do well in other professional workloads". They would generally test different workloads, and relative performance can and does vary depending on the specific software used. I'll take Cinebench scores, but I also keep in mind it is the best case scaling of any benchmark that is commonly run.

[Taf the Ghost]

1 minute ago, porina said:

It isn't an area I've looked into closely, but what would be a "professional workload"? I don't recall any reputable reviewer saying words to the effect "It runs Cinebench great, so it'll also do well in other professional workloads". They would generally test different workloads, and relative performance can and does vary depending on the specific software used. I'll take Cinebench scores, but I also keep in mind it is the best case scaling of any benchmark that is commonly run.

Blender, Handbrake, C-Ray, modeling programs and most of the editing programs. R20, especially, is a good combined workload that threads well. It also lets you "sanity check' results pretty well. It's very useful benchmark to have publicly available in general, but Cinema 4D is a program that's used professionally.

 

If you want deeper, individual program benchmarks, hit up Puget Systems. They've got a lot of them.

[ravenshrike]

31 minutes ago, Taf the Ghost said:

Once 7nm Zen2 APUs are out, I expect all of those programs to be in AMD's favor.

WinRAR probably won't be, but that's because WinRAR is either a poor program or poor compression algorithm. Has anyone done RAR (de)compression tests in 7Zip to find out?

[CiBi]

1 hour ago, leadeater said:

I leave my TC and ESC on always, not looking to buy a new car . But with that all on 0-60 in 5.3. Not looking to tear through BS RE050A's quickly either, too expensive.

What car are you driving?

[Taf the Ghost]

Just now, ravenshrike said:

WinRAR probably won't be, but that's because WinRAR is either a poor program or poor compression algorithm. Has anyone done RAR (de)compression tests in 7Zip to find out?

https://www.techspot.com/review/1869-amd-ryzen-3900x-ryzen-3700x/

 

See the WinRAR and 7Zip results.

[Touch My Hamm]

Kinda silly on Intels part. Do they watch the reviews? Yes there is usually a segment with Cinebench since it can be used as a standard baseline where its more of a level playing field. But alot of reviews then show in game benchmarks. This usually has Intel in the lead since they have higher clock speeds on single threads and have drivers for each game. This can heavily skew data just having game benchmarks since when a reviewer looks at the CPU it could be using prerelease drivers which could have a negative impact on performance. 

 

Its silly of Intel to ask reviewers to change what they are doing since its the consumer who likes to see cinebench scores. instead of fighting against what their consumers are doing why not work with them? Its about time that Intel had to actually work for their position as the "Top gaming cpu's".

[porina]

4 minutes ago, Taf the Ghost said:

Blender, Handbrake, C-Ray, modeling programs and most of the editing programs. R20, especially, is a good combined workload that threads well. It also lets you "sanity check' results pretty well. It's very useful benchmark to have publicly available in general, but Cinema 4D is a program that's used professionally.

Thanks for the clarification.

 

When people are buying a system for specific task(s), for sure it'll make sense to get something with good performance on that, but it becomes more of a narrow focus than the broader landscape.

[5x5]

30 minutes ago, porina said:

Thanks for the clarification.

 

When people are buying a system for specific task(s), for sure it'll make sense to get something with good performance on that, but it becomes more of a narrow focus than the broader landscape.

Intel were talking about HEDT, though. If those aren't the broad landscape for 12+ cores, I don't know what is.

[corsairian]

Can we go back to testing Crysis as the one benchmark to rule them all?

[pas008]

1 hour ago, porina said:

Different types of code will behave differently, and be influenced by the rest of the system differently. Cinebench R15 has a characteristic of scaling near perfectly with cores, has a high benefit from HT/SMT, and is minimally affected by ram performance. Haven't got to know R20 as well yet, but it seems pretty similar. It is near enough the ideal case to show off peak CPU performance. 

 

this is the main problem here people keep forgetting

isa, ipc, frequency, memory, cores, compilers etc all play roles

so many variables that these synthetic benchmarks are nice to get an ideal and sterile but no way real

 

pcs are multitasking machines and in no way anyone simply runs these alone with everything else closed, which I think reviewers should include multitasking bench scenarios with real life uses for few similar products

 

 

 

[porina]

1 hour ago, 5x5 said:

Intel were talking about HEDT, though. If those aren't the broad landscape for 12+ cores, I don't know what is.

OP says "notebook and 2-in-1 devices"

[Lathlaer]

18 minutes ago, porina said:

OP says "notebook and 2-in-1 devices"

Exactly, I wonder how high Steam and LoL are amongst x299 and TR4 users

[5x5]

35 minutes ago, porina said:

OP says "notebook and 2-in-1 devices"

That's what Intel are basing their review "suggestion" for HEDT parts on.

[Master Disaster]

2 hours ago, 5x5 said:

That's what Intel are basing their review "suggestion" for HEDT parts on.

Nah, intel are using data collected on portable devices to show how the average DESKTOP user uses their PC.

 

That's disingenuous at best and flat out misleading and misrepresentation at worst.

 

It's like BMW collecting speed data from the M5 then releasing a marketing slide saying "all our cars can do 180mph" with an annotation saying "data collected from the M5 only".