Ryzen and memory

[chew*]

Forgive me but I remember hearing that the Bios was embeded on the CPU, correct? How does manufacturer baord bios/firmware fit in all this?

I think that's grey area for NDA so I will have to wait to give you the answer.

[dasa]

Those memory benchmarks are perfectly feasible as i know that 20%+ scaling is possible (depending on timings) and the witcher does start getting cpu limited around 100fps in a few parts of the game
digital foundry has shown ~8% improvement from 2133 to 3000 in it
my guess is that the funny scaling at different speed steps could be timing related or just margin of error in there test
for example they may be using 2133c13 2400c15 2666c16 3000c15 3200c14

[chew*]

Yah the timings are key. without knowing those it leaves a rather large variable in the equasion.

If timings were same through all tests? that scaling could be realistic but also dumb.

[dasa]

intel exactly matching amd on a number of the points though is more than a little unlikely
almost like they mixed up there tests a bit

[BenchZowner]

Without knowing timings used its a shot in the dark to make assumptions. if identical timings were used its realistic however. There is also an X factor that I'm not at liberty to discuss yet.

Grid autosport is very sensitive to total platform performance albeit a little dated. Its also real world vs synthetic.

I'm likely going to use it as a reference in testing since it is highly accurate and consistent and does not require me to make my own custom benchmark.

It also allows other users to do a direct comparison of there own system versus my results which would be impossible with a custom made benchmark unless it was hosted for public use.

If you are talking about GRID Autosport's integrated benchmark, I'd say quite the opposite, it's far from consistent.
5 runs with the same config 1 after the other and the results are over the place even in the highly unlikely to change a lot average fps.

manual ( real ) Gameplay would be consistent if you can keep your driving and route consistent, however I was simply bored and lacking in time to do manual tests apart from BF1 and Watch Dogs 2 for the current tests.

[NEOAethyr]

I think that's grey area for NDA so I will have to wait to give you the answer.

I'm concerned a bit about this.
That and the so called "restricted RAM tuning options".
Makes me wonder if you'll kill the cpu if you push the ram to much.
What would normally be fixable with a cmos reset or a bios flash is stuck on the cpu.

I hope I'm reading to much into it.
I suppose even if there was an issue I guess the bios makers could just limit the min and max timing mask.

...
Other then that I am looking forward to what can be done with these new ryzen chips .

Oh and I've been reading that there's a new microcode coming out a while after launch that should improve mem clocking?
Microcodes normally don't do that..., but it sounds like some sort of patch.
I wonder if the asus has flashback...

[chew*]

I can't see them not adding them in the future. It only benefits them if users can extract a tad more performance.

flashback on the asus boards that I'm aware of is done by using rog connect and renaming file. no rog connect then I'm not sure.

[cdawall]

I have so many things I want to say... Currently using a 3600 kit that seem to be doing ok.

[chinook]

Good things?

[robbicfc]

Seems AMD is also making sure that it is known that Ryzen 7 does not mind faster than DDR4 3200 - See Wccftech - R7 1700 with G Skill DDR4 3400 on Asus Crosshair Hero

[cdawall]

Good things?

We need a more complete bios is the only current issue I am having. For something hitting the market in a couple days it seems very unfinished to me.

[chinook]

I see. Do you have info that it affects all .. players? Or just this particular one that you have? And thanks for the info, mate.

[cdawall]

I can't speak for anyone other than me and I have a single board.

[Kal-EL]

If you are talking about GRID Autosport's integrated benchmark, I'd say quite the opposite, it's far from consistent.
5 runs with the same config 1 after the other and the results are over the place even in the highly unlikely to change a lot average fps.

manual ( real ) Gameplay would be consistent if you can keep your driving and route consistent, however I was simply bored and lacking in time to do manual tests apart from BF1 and Watch Dogs 2 for the current tests.

BZ u benchin once more? nooooooooice

[BenchZowner]

Got like 4 hours of benching video to put some "high speed RAM, more cores = better performance" claims to the test

[Ket]

Just a FYI guys that my review and testing will be late to the party. I have the CPU but the shipment of the Auros Gaming 5 is delayed - Gigabyte falling behind apparently. So my review won't come until the end of next week at the earliest

AMD wanted to calm people down a bit ;-)

https://twitter.com/AMDRyzen/status/836285599636127744

Timings are weak, but its kinda reasurring

On the bright side, thats with a 1T command rate at least. Still a bit rubbish though.

[chew*]

For now i would pass on the corsair platinum 10-12-12-28 kit. I got them to post and boot and run 32m but they defaulted at 2T CR.

It required doing the gigaboo gigabyte rain dance reboot cycle to get them working.

They are also double sided.

[cdawall]

This is the kit I am using

http://www.corsair.com/en-us/vengean...6gx4m2b3600c18

[chew*]

Found a better way to get them to boot. Used reviwers kit to get settings applied. Did the old swaparoo trick after shutdown. No reboot dance.

Could not fool CR though....back at 2T

[eagle101]

Found a better way to get them to boot. Used reviwers kit to get settings applied. Did the old swaparoo trick after shutdown. No reboot dance.

Could not fool CR though....back at 2T

what no 1T.....

[chew*]

I would probably opt for the gskills so far eagle. I dunno how but at 2T (unless not relevant anymore ) in one test is comparable and that test seems inconsistant in W10x64. The rest its not even close.

To bad corsairs website fails to mention 2T.

I would not have ordered.

Hell the budget vengeance have 1T

[eagle101]

I would probably opt for the gskills so far eagle. I dunno how but at 2T (unless not relevant anymore ) in one test is comparable and that test seems inconsistant in W10x64. The rest its not even close.

To bad corsairs website fails to mention 2T.

I would not have ordered.

Hell the budget vengeance have 1T

at least progress is being made.......... and ya I hope the G.Skill Trident-Z might work excellent no body knows about that one yet unless someone out there
has tried them?

[chew*]

The trident z is what im using in pc 3200 14-14-14 flavor.
Samsung

Corsair in 15-17-17 pc3000
Hynix low bin

Corsair in 10-12-12 pc2400
Hynix i dont know bin....

[Hell Hound]

What about 4 sticks ?

[chew*]

What about 4 sticks ?

You gonna buy them? .

No offense but not spending $300 to make a post on a forum if it boots or not.

I'll let someone else who has them already confirm it. Im chasing platforms performance. High density and high performance usually dont mix.

I game more than 16g is pointless. Running 4x4 is dumb not to mention options are slim to none.

Double sided being finicky is a good indicator of 4 dimm drama and 2T CR at this point in time.

Bios,agesa could change. Stuffs new and has some teething issues.