[Hardware Unboxed] R7 1700 @4Ghz + 3200Mhz RAm = ~7820X @4.5Ghz

[Anarchist Mae]

My problem is when a review pairs Ryzen with 2666Mhz RAM to compare it against a competitor running 3600Mhz RAM, and then pointing out the poorer gaming performance.

Or if they reuse launch day benchmarks months later when comparing to a new CPU release.

 

[Zucker2k]

My problem is when a review pairs Ryzen with 2666Mhz RAM to compare it against a competitor running 3600Mhz RAM, and then pointing out the poorer gaming performance.

Show us where 3600mhz RAM significantly impacted results for SKL-X in gaming.

 

[Anarchist Mae]

Show us where 3600mhz RAM significantly impacted results for SKL-X in gaming.

That isn't the problem that's being highlighted.

The problem is that the CPU that does not benefit fast ram is being compared to a CPU that does benefit from fast ram, except they give the faster more expensive ram to the system with the CPU that does not benefit from it.

Why does it seem to be up to everyone else to do your thinking for you? I'm getting really sick of arguing with you because you don't seem to read what others are saying and instead interpret what they say in a way that suits you. That's called making a straw man argument by the way.

 

[Zucker2k]

That isn't the problem that's being highlighted.

The problem is that the CPU that does not benefit fast ram is being compared to a CPU that does benefit from fast ram, except they give the faster more expensive ram to the system with the CPU that does not benefit from it.

Why does it seem to be up to everyone else to do your thinking for you?

If you can't show that the competitor chip benefitted from the faster rated RAM then what's the problem? 3600mhz RAM on Ryzen involves some research. Even the link in the OP only shows 3200mhz. I don't even know if you can run 4 sticks at 3600mhz speeds on Ryzen. Care to educate me on the intricacies of this subject?

 

[Anarchist Mae]

If you can't show that the competitor chip benefitted from the faster rated RAM then what's the problem? 3600mhz RAM on Ryzen involves some research. Even the link in the OP only shows 3200mhz. I don't even know if you can run 4 sticks at 3600mhz speeds on Ryzen. Care to educate me on the intricacies of this subject?

It's pretty unlikely to get even 4 sticks of 3200MHz ram working on Ryzen, and I very much doubt 3600MHz ram would work either. Of course I can only go on what other people have said, and I'm not sure where the latest bios update leaves things.

Personally, I think it's the job of the reviewer to know these things, after all who better to educate us on the possibilities of our hardware? And since Ryzen does benefit from faster ram, if not in that exact configuration, it's hardly a fair $/performance comparison.

Edit: I mean that both ways btw, it isn't fair on the more expensive CPU to pair it with with the expensive ram that it cannot use and drive the total price up for no good reason.

 

[Crumpet]

If you can't show that the competitor chip benefitted from the faster rated RAM then what's the problem? 3600mhz RAM on Ryzen involves some research. Even the link in the OP only shows 3200mhz. I don't even know if you can run 4 sticks at 3600mhz speeds on Ryzen. Care to educate me on the intricacies of this subject?

You can't, but you can run 2 sticks at 3600Mhz.

As Stoatally says, its up to the reviewers to provide a fair testing scenario for the benefit of the reader. A fair testing scenario is important, it's called science.

Yes, Ryzen and its varieties have some issues with slower speed RAM, and Intel has a significant advantage there, undoubtedly. But that doesn't give reviewers the right to show skewed and unfair benchmarks and claim them as fact. But what is also fact, is that Ryzen has incredible gains with faster RAM, and whilst the 7820X does make improvements with the faster RAM, Ryzen undoubtedly gains on it quite significantly.

 

[Crumpet]

Anything you want. We can start with Cinebench, that way your R9 290 won't slow you down...

Unfortunately new RAM wasn't playing ball at 2666, oddly enough, but purrs along happily at 3600 CAS 16.. whodathunkit.

Anyway, my gaming suite is probably completely defunct apart from GTA V and we both know i'm going to lose that one, so I compiled a small spreadsheet of CPU bench results for you;

-edit- Worth noting that 3.7Ghz means stock cpu, 4.1Ghz turbo with 3.7Ghz all core.

This is on a cheap SSD 120Gb boot drive with 2 storage drives for games and other crud, I have many programs installed and have CAM running in the background to control my fans, plus windows defender, radeon settings, sonic suite 3 and the realtek HD audio manager. So this is hardly a clean testing rig.

https://docs.google.com/spreadsheets/d/1AUwtX-UkPfRqsDAJlP4DDl_mrCG77z-NIRrYrpdlDdA/edit?usp=sharing

Spoiler: Cinebench

Geekbench4@Stock
Geekbench4@4Ghz
Geekbench4@4.1Ghz

 

[IEC]

If you can't show that the competitor chip benefitted from the faster rated RAM then what's the problem? 3600mhz RAM on Ryzen involves some research. Even the link in the OP only shows 3200mhz. I don't even know if you can run 4 sticks at 3600mhz speeds on Ryzen. Care to educate me on the intricacies of this subject?

You can run 4 sticks of Samsung B-die at up to 3600 CL16, assuming good RAM. Speeds beyond 3466 generally require tweaking of subtimings.

Memory compatibility and overclockability has generally been improving with each UEFI update.

 

[Crumpet]

You can run 4 sticks of Samsung B-die at up to 3600 CL16, assuming good RAM. Speeds beyond 3466 generally require tweaking of subtimings.

Memory compatibility and overclockability has generally been improving with each UEFI update.

I believe very, very few people have hit over 2933 with 4 rank though, but it is slowly getting there yes.

 

[IEC]

I believe very, very few people have hit over 2933 with 4 rank though, but it is slowly getting there yes.

I was hitting 4x8GB, 3200 CL14 by AGESA 1.0.0.4. It wasn't plug and play, but it was pretty easy.

With AGESA 1.0.0.6, 3200 CL14 is plug and play. Needs more tweaks to hit 3466+

 

[Anarchist Mae]

I was hitting 4x8GB, 3200 CL14 by AGESA 1.0.0.4. It wasn't plug and play, but it was pretty easy.

With AGESA 1.0.0.6, 3200 CL14 is plug and play. Needs more tweaks to hit 3466+

At most I'd expect reviewers to update their bios and enable the XMP profile for their ram, anything above and beyond that while great is not something the average home user is going to be comfortable with.

 

[IEC]

Got curious what I could do on my cheapest Ryzen rig (not in sig yet, just built). Surprisingly XMP2 profile worked out of the box for some DDR4-3600:

Vital specs:
Ryzen 1600X ($200 on sale)
MSI B350M Gaming Pro ($70 mobo)
Galax HOF 2x8GB kit DDR4-3600 CL17 (ostensibly B-die, $120 on sale)

Ignore CPU temps, literally had small aluminum block gravity mounted for testing purposes.

 

[TahoeDust]

Unfortunately new RAM wasn't playing ball at 2666, oddly enough, but purrs along happily at 3600 CAS 16.. whodathunkit.

Anyway, my gaming suite is probably completely defunct apart from GTA V and we both know i'm going to lose that one, so I compiled a small spreadsheet of CPU bench results for you;

-edit- Worth noting that 3.7Ghz means stock cpu, 4.1Ghz turbo with 3.7Ghz all core.

This is on a cheap SSD 120Gb boot drive with 2 storage drives for games and other crud, I have many programs installed and have CAM running in the background to control my fans, plus windows defender, radeon settings, sonic suite 3 and the realtek HD audio manager. So this is hardly a clean testing rig.

https://docs.google.com/spreadsheets/d/1AUwtX-UkPfRqsDAJlP4DDl_mrCG77z-NIRrYrpdlDdA/edit?usp=sharing

Spoiler: Cinebench

Geekbench4@Stock
Geekbench4@4Ghz
Geekbench4@4.1Ghz

Here you go. The only surprise was the multicore CPU-z test...closer than it seems like it should be...

https://browser.geekbench.com/v4/cpu/3600868

Firestrike Physics is: 25 618
http://www.3dmark.com/fs/13233939

 

[Crumpet]

Smoked me in GB4, that seems like too much of a gap to me, 2k single and 12k multi!

~3k difference in Firestrike physics I could believe.

-Edit- hang on, 3.6Ghz. No way. No chance.

 

[Osjur]

Crumpet would you be willing to run 3dmark API draw call test with that 3600mhz CL16 and 4ghz? I wanna see what kind of difference it makes compared to 3466mhz CL14 and tightened subs.

 

[Crumpet]

Crumpet would you be willing to run 3dmark API draw call test with that 3600mhz CL16 and 4ghz? I wanna see what kind of difference it makes compared to 3466mhz CL14 and tightened subs.

You need to purchase 3dMark for that if i'm correct??

If not for sure, if so.. i'll see how much it is.

 

[TahoeDust]

Smoked me in GB4, that seems like too much of a gap to me, 2k single and 12k multi!

~3k difference in Firestrike physics I could believe.

-Edit- hang on, 3.6Ghz. No way. No chance.

Nah...that is reporting incorrectly. They were all run @ 5.0GHz.

 

[Crumpet]

Nah...that is reporting incorrectly. They were all run @ 5.0GHz.

5Ghz, on a 7820X? jesus. Delidded etc i'm guessing?

Even so, picking apart the scores GB4 seems to HEAVILY favour Intel. A friend is testing his 6600k @4.2 with 2133Mhz RAM and it scores 600 more in Single core and has about 65% of the multicore score.. 4 threads vs 16, only 100Mhz faster clock with vastly slower RAM.

 

[TahoeDust]

5Ghz, on a 7820X? jesus. Delidded etc i'm guessing?

Even so, picking apart the scores GB4 seems to HEAVILY favour Intel. A friend is testing his 6600k @4.2 with 2133Mhz RAM and it scores 600 more in Single core and has about 65% of the multicore score.. 4 threads vs 16, only 100Mhz faster clock with vastly slower RAM.

Yeah...it is delidded. I could bench it at 4.9GHz before delidding. It runs 4.8GHz 24/7.

 

[Crumpet]

Yeah...it is delidded. I could bench it at 4.9GHz before delidding. It runs 4.8GHz 24/7.

Thats an insanely high clock, nice god chip. Mines pretty good for a Ryzen but damn.

Even so. I'm not sure why GB seems to score the 7820X so absolutely massively favourably.

 

[Osjur]

You need to purchase 3dMark for that if i'm correct??

If not for sure, if so.. i'll see how much it is.

Sent you a PM

 

[TahoeDust]

Thats an insanely high clock, nice god chip. Mines pretty good for a Ryzen but damn.

Even so. I'm not sure why GB seems to score the 7820X so absolutely massively favourably.

It will bench single core @ 5.1GHz.... Ignore the voltage in CPU-z...it is reporting incorrectly.

 

[Zucker2k]

Yeah...it is delidded. I could bench it at 4.9GHz before delidding. It runs 4.8GHz 24/7.

Hotdamn! How are you cooling that monster?

 

[kush120]

@TahoeDust

What was the budget for your Skylake-X rig?

Once my TR build gets here, We can compare $ for $.

Currently running Ryzen 7 3.9ghz, 3333mhz CL14 memory, 960evo, GTX 1070 . Gonna be picking up the 16core Threadripper in a couple weeks tho.

 

[Osjur]

Hotdamn! How are you cooling that monster?

You can do it with good custom loop and delid with LM, if you don't need to apply too much voltage. And its "only" 8 cores.