Prebuilt [SFFn] ASRock's DeskMini A300 - Finally!
- Thread starter confusis
- Start date
based on pro 4650g, you should not have any problems with it and the only performance difference could be memory latency (tsme - amd memory encryption, if it applies to picasso as well).
pro 3400g is not listed on a300 official cpu support page though, but 2200/2400g pros are there ?
@DoM_H Check your retailer. There is no PRO model of 3400G. You're right. Even AMD lists PRO on their product page.
What has PRO what the non-PRO does not offer? It's "AMD Memory Guard" and "AMD GuardMI Technology"
If rubicoin is able to run with his 4650 PRO, I suggest you would be able to use 3400G PRO in the same way.
@rubicoin impressive benchmarks! could you add it up with a small GPU review? So we could tests both Vega's against each other?
I can test against UNIENGINE Valley, CSGO, GTA V
We should speak about the testing profiles though, and mem timings
UNIENGINE Valley is free, I've achieved 45fps or ~1900 points with:
API: DX11
Quality: Low
Stereo 3D: Disabled
Monitors: Single
Anti-aliasing: Off
Full Screen: Yes
Resolution: 1920x1080
at stock Ryzen 2400G with G.Skill 3200MHz@CL18 XMP.
CINEBENCH R15: 775 points
CINEBENCH R20: 1713 points
What has PRO what the non-PRO does not offer? It's "AMD Memory Guard" and "AMD GuardMI Technology"
If rubicoin is able to run with his 4650 PRO, I suggest you would be able to use 3400G PRO in the same way.
@rubicoin impressive benchmarks! could you add it up with a small GPU review? So we could tests both Vega's against each other?
I can test against UNIENGINE Valley, CSGO, GTA V
We should speak about the testing profiles though, and mem timings
UNIENGINE Valley is free, I've achieved 45fps or ~1900 points with:
API: DX11
Quality: Low
Stereo 3D: Disabled
Monitors: Single
Anti-aliasing: Off
Full Screen: Yes
Resolution: 1920x1080
at stock Ryzen 2400G with G.Skill 3200MHz@CL18 XMP.
CINEBENCH R15: 775 points
CINEBENCH R20: 1713 points
Last edited:
@DoM_HCheck your retailer. There is no PRO model of 3400G.You're right. Even AMD lists PRO on their product page.
What has PRO what the non-PRO does not offer? It's "AMD Memory Guard" and "AMD GuardMI Technology"
If rubicoin is able to run with his 4650 PRO, I suggest you would be able to use 3400G PRO in the same way.
@rubicoin impressive benchmarks! could you add it up with a small GPU review? So we could tests both Vega's against each other?
I can test against UNIENGINE Valley, CSGO, GTA V
We should speak about the testing profiles though, and mem timings
UNIENGINE Valley is free, I've achieved 45fps or ~1900 points with:
API: DX11
Quality: Low
Stereo 3D: Disabled
Monitors: Single
Anti-aliasing: Off
Full Screen: Yes
Resolution: 1920x1080
at stock Ryzen 2400G with G.Skill 3200MHz@CL18 XMP.
CINEBENCH R15: 775 points
CINEBENCH R20: 1713 points
my initial quick tests
- 3dmark time spy score: 1362 points (1198 graphics and 6201 cpu score), very unimpressive fps values (7-8 fps graphics and 20 fps cpu)
- gaming: only played fallout3 nv @ maxed fullhd ultra settings, stable 60 fps (freesync=on, locked it to 60 i guess) with occasional frame drops.
- memory: ripjaws run stable @ 3200 now (1600 mhz fclk), did not try raising it yet. already awesome as with 2200g it was not stable even at stock 3000 mhz.
- just ran unigine valley for you (looped for 15 min, did not realize it has a separate benchmark button, lol). results with same ram speed and settings as yours (windowed only because of ultrawide screen):
@rubicoin thank you very much! this sounds nice! the new CPU yields very nice results. About Vega, I am dissapointed.
The 3.60R BIOS (vs. AGESA v1) gave me +5fps more in Valley leading to ~45 fps. Now, you have 2 frames more. But it could be also error margin.
CPU almost twice as fast (worth of an upgrade) / Vega is "almost" the same... Well be observing the 47x0G's Vega.
Yea, Valley is a 18-scenes-loop until you hit "Benchmark" yea, sry about that, I forget it still sometimes
The 3.60R BIOS (vs. AGESA v1) gave me +5fps more in Valley leading to ~45 fps. Now, you have 2 frames more. But it could be also error margin.
CPU almost twice as fast (worth of an upgrade) / Vega is "almost" the same... Well be observing the 47x0G's Vega.
Yea, Valley is a 18-scenes-loop until you hit "Benchmark"
yea, sry about that, I forget it still sometimesYour temps are higher than mine at idle. Im sitting at 29 with the stock wraith cooler while browsing through SFF. Dont know what your ambient temps are but it cant make that much of a difference.
I have the same cooler, do you mind to tell me how to mod this cooler for using 25 mm fan?
25 °C ambient. but i did a mistake with these values: all temps reported are CPUTIN=motherboard cpu sensor values as no core temp is displayed in current hwmonitor version! i did not realize it first, but these can be completely false numbers, sorry. i don't think that i have thermal problems, applied noctua nt-h1 paste with care and fastened all screws properly, so I'll check temps later with adrenaline or some other software.
for the record, 4650g starts at CPUTIN=33 °C and slowly goes up to 36+, and no core/package temp rows:
(after 15 min cpu-z stress test @ 2000 rpm - way too load for me! - CPUTIN went up to 59 °C)
yosoywilson78: what are the idle/load rpms of wraith cooler for you? what is your cpu, 4650g as well?
A300: about 15 mm > 25 mm fan mod, see my posts about it here. one thing to note: after opening/closing case a couple of times, disassembling became much easier and you can lift the lid a bit by hand and pull out the tray without removing side screws.
for the record, 4650g starts at CPUTIN=33 °C and slowly goes up to 36+, and no core/package temp rows:
(after 15 min cpu-z stress test @ 2000 rpm - way too load for me! - CPUTIN went up to 59 °C)
yosoywilson78: what are the idle/load rpms of wraith cooler for you? what is your cpu, 4650g as well?
A300: about 15 mm > 25 mm fan mod, see my posts about it here. one thing to note: after opening/closing case a couple of times, disassembling became much easier and you can lift the lid a bit by hand and pull out the tray without removing side screws.
Last edited:
aida64 temps are far better for comparision i guess:
2200g had 29 °C idle / 64 °C stress with same colling (hwmonitor package temps). not sure which aida temps to compare to (cpu vs cpu diode).
3.62 bios file date is 22/01/2020, while 3.60R beta is 14/08/2020 ?
checked 4PDA thread with google translate, found nothing related.
checked 4PDA thread with google translate, found nothing related.
While it is true the file has been modified on the 1st if january. It seems like it has been created on the 13/08/2020.
So the creation date is after the modification date ?
Well, let's wait for a brave explorer (or a release note).
it's possible that 3.62 is the next stable version (just not officially released yet), based on 3.60R which would be the last beta then. it'd make sense as 3.60R works great with 4650g as well.
without release notes I'd rather not try it. if it was an old version without renoir support, i'd be screwed. let's wait for a JZ version
Just wanna share some statistics I gathered with my A300 with 2400G / 4750G and notebook 4650U
This A300 + 4750G is perfect for small VM applications, especially in 35W TDP config, <60C in small FFT is just insane.
This A300 + 4750G is perfect for small VM applications, especially in 35W TDP config, <60C in small FFT is just insane.
yeah, probably, no GnbSocRenoirDxe string (or any other with "Renoir" in name like in 3.60N+) in UEFIextract report ?
Last edited:
thanks for the reference, know i have something to match my temps (35w 4750g vs 65w 4650g seems to be a fair comparison), and also know which is the best software to use. the only question remaining is your cooling solution.
so i consider this hwinfo report the most accurate measuring possible (28 °C idle and 60 °C stress temps):
I am using a Noctua L9a with A9x25 on my A300
And this is the Temps during stress test for 4750G 65W configuration
Similar threads
