Ryzen - Return of the Jedi

[chew*]

Motherboard im using and motherboard vendor rants in general.

This board is underpowered first off. Its also a gaming 5. theres already a 7 on the site. I hope it does not share the same pwm as this one.

This board has no clock gen chip therefore bclock is locked. Who cares right? wrong. In order to extract the most performance from Ryzen you will need to clock bclok up past 100. Max memory multipler is 32 and I can not confirm but in cpu-z NB is always equal to ram. If indeed that is NB frequency to extract even more performance will require us to push the memory to beyond 3200 to get the northbridge frequency up.

Boards in general. I was benching 3d game tests at 4.2 gig. I could not pass cpu tests at 4.2. Had the board had a simple tweak it or msi oc knob or any type of button to dial the bclock frequency up during game tests and down during cpu tests I could have gotten a better result.

Unfortunately vendors have decided to screen print boards add lights and install heat trapping devices over PWM sinks that don't even adhere to the simple law of HEAT RISES.

UPDATE
Lets rewind a little. 990FX, now I can understand that BD sucked ( yes I can say that now ) and the board could handle thuban Deneb just fine but lets look at the heat dissapation qualities.

Not bad not absolutely magical but not bad......Asus probably had the most sensible solution here with crisscross finned PWM sink.
http://chew.ln2cooling.com/?Qwd=./blah&Qif=IMG_3751.jpg

How about crosshair 3?
http://chew.ln2cooling.com/?Qwd=./Cr...f=IMG_1652.JPG

MSI 890
http://chew.ln2cooling.com/?Qwd=./MS...f=fullshot.JPG

gigabyte 890
http://chew.ln2cooling.com/?Qwd=./89...iga%20comp.jpg

gigabyte 790
http://chew.ln2cooling.com/?Qwd=./955&Qif=IMG_1274.jpg

As we can see all these had common sense heatsinks that served a functional purpose. dissapate heat.

END

Im very dissapointed and when i was working with vendors i would reccomend at least a little common sense and funtional features that helped overclockers.

To be continued. I need some sleep. I will rattle off whatever bugs i forgot this board has when i get up and workarounds i have found to get around them.

Gigabyte G5

bclock not 100. no workaround up to bios team.

SMT disabling bug. When enabling SMT again posting and going to windows don't go Oh @#$% I think i killed my chip when threads do not come back in windows. simply shut down, power off psu till power light goes out. switch back on and boot.

UPDATE
New workaround for SMT disabled core 0 bug with 1700 non X cpu.

Here is what the bug looks like booting with OC from bios.



WORK AROUND

CPU ratio AUTO
Set everything else in bios as you would normally including voltages required for your known stable target cpu clock.
Boot. Once on desktop open Ryzen Master. set all core multi to desired frequency. verify all cores is cpu-z are at target frequency.
Yes I know this annoying but it works for now till we can fix it via multiplier from bios.



Reported. Pending bios update

END

General gigabyte post cycles. Try not to change to many settings at once. example c and q off smt off set ram speed and tune timings. Do one at a time save reboot change next. for as long as i have used gigabyte boards they still do a funky reboot cycle dance. This gets around that unless you have a serious problem. I have gotten bclock up on boards that have hated bclock changes by doing this ( p67 ) all the way up to 110. 1 step up reboot. another step reboot.

Update
ALL VENDOR cpu-z bug. SMT disabled shows only 4c/4t tabbing the cores shows two physical and two logical. Workaround = dont use vendor cpu-z. The normal cpu-z reports it properly.



END

More info to come but i need sleep