Welcome Guest ( Log In | Register )

17 Pages « < 2 3 4 5 6 > » Bottom

Outline · [ Standard ] · Linear+

 AMD Ryzen, AM4 / AM5 Platform

views
     
nrw
post Sep 3 2019, 07:40 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(PedangGila @ Sep 3 2019, 07:13 PM)
My 3600 I think never get to 4.2. The best it can do is 4.1 but with 1.450v which is quite alarming. Stock cooler. I dont want to tinker with OC setting as PC sometimes refuse to boot, requiring manual power kill to do so. Still on 1A bios for this b450 tomahawk.

ANy of you experience 'normal' performance with bios 1B version?
*
for boosting behaviour there's not much difference but for memory setting the B4 bios was quite an improvement.
so it won't hurt trying it out. maybe you get lucky and boosting changes.
what does your single core / all core boost to when running cinebench r15 for example?
nrw
post Sep 4 2019, 08:46 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


Interesting take from 1usmus

QUOTE
Derbaur loves pulling a blanket over himself. There is one problem - the labeling of cores. Almost 3 months ago, I notified AMD that most processors will not reach their maximum boost for a number of reasons. The update that the community will receive in a week partially solves these shortcomings.

https://twitter.com/uzzi38/status/1169142110223773696


I will explain in brief what is happening. ASICs for each core are defined at the factory. The goal of successful cores is to achieve a certain frequency within the FIT, which is in the region of 1.487 volts for a low current load. The stability of the system is monitored by Clock Stretcher technology and AVFS, which automatically adjusts the real core frequency depending on the information received from dozens of sensors. If we try to load an unsuccessful core - it will not get the maximum frequency, we will get a bottleneck by current, voltage or temperature. Most processors of the Zen + and Zen 2 generation have non-optimal core markings.
As an example, I will show you the results of testing my own instances of 2700X, 3700X and 3900X.
I like his take on derbauer, too. That guy has to many financial involvements and to little statistical knowledge to be taken for granted on every occasion.
nrw
post Sep 9 2019, 06:25 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(dRLurve @ Sep 9 2019, 05:50 PM)
soso deducting the 3600 that's ~925 for the vii hero which is about 800 in good used condition.
nrw
post Sep 9 2019, 06:52 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(dRLurve @ Sep 9 2019, 06:41 PM)
Bro, Can elaborate more? What your mean good used condition?

Checked the recent price list the cpu alone close to RM850 and Crosshair at 1.2K?

This pwp only priced at 1.7k. Saved RM300?

Do recommend if you have better bargain. Thanks.
*
lets put it like that, if you insist to buy new motherboard its an OK/good price.
but for a 3600 the vii is pretty overkill already, maybe you wanna eye the 3700x?

even though the price tag for the c7h is 1.2k new, there would be better options (cheaper x470/b450 or x570) around probably before one would consider getting it at that premium price tag.

This post has been edited by nrw: Sep 9 2019, 06:54 PM
nrw
post Sep 10 2019, 09:21 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


seems latest ABBA rclxms.gif "Mamma mia, here I go again whistling.gif " agesa & 46.49 SMU still don't fix max boost

This post has been edited by nrw: Sep 10 2019, 09:24 PM
nrw
post Sep 11 2019, 09:46 AM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(AMDAthlon @ Sep 10 2019, 10:04 PM)
Still on non-official BIOS right? Some of the reddit users using Beta ABBA BIOS already reporting it achieved boost clock. For now better wait for AMD official response..
*
the smu is official, and despite current tests rely on beta bios', what they did is simply add the smu to an abb bios.

seems like for anything below the 3900x this has improved single core boosts.

nrw
post Sep 11 2019, 11:17 AM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(TristanX @ Sep 11 2019, 11:03 AM)
Motherboard makers also needs to do internal testing too.

Gigabyte is quiet so far.
*
Gigabyte kinda quickest tongue.gif

https://forums.tweaktown.com/gigabyte/28441...-beta-bios.html

Not that I'd encourage you to test, especially with your 3900
nrw
post Sep 12 2019, 12:14 AM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(vernon91 @ Sep 11 2019, 10:01 PM)
Well my RAM model wast in the QVL list but from what i have checked all A-DATA ram in the QVL supports 4 DIMM config

+

some other brand RAM in the QVL list that also uses the same H5AN8G8NCJR-VKC (Hynix CJR) DRAM Part Number supports 4 DIMM config as well.  hmm.gif
*
ye I wouldn't worry to much with zen2.
even with daisy chain boards 4 sticks will run easily until at least 3600, most board will even do 3800.
and that's all you need as there is hardly any point to sacrifice 1:1 for higher frequencies.
nrw
post Sep 20 2019, 12:12 AM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


November launch: 3500x
6c6t. Will be very nice option for those who really only do gaming (and are on a tight budget.)
nrw
post Sep 20 2019, 12:25 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(John Chaser @ Sep 20 2019, 12:02 PM)
Isn't the cache rumored to have been reduced to 16MB on this?
*
I don't think so, they'll just take 3600's and disable smt

caches are fixed per number of cores..for now.
nrw
post Sep 20 2019, 07:43 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(hwchua @ Sep 15 2019, 03:33 PM)
AMD Ryzen 5 3400G  (There is 2nd and 3rd Generations ... any different? Any advise, thanks
*
2200g -> 3200g (slight OC, 6 instead of 4MB total cache)
2400g -> 3400g (slight OC, 3400g should be soldered somemore for better temps, 6 instead of 4MB total cache)
nrw
post Sep 21 2019, 01:08 AM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(shirohamada @ Sep 20 2019, 11:51 PM)
L3 cache is not.
They can fuse it off and it does take die area. But if yield is good, they'll just turn off smt.
*
so far:
... l3 cache is fixed to 32 per ccd
... l2 cache is fixed to .5 per core

for 6 cores and 4.1 you don't even need good yield, that's lowest tier.
maybe they bring a 3500 non x and deactivate l3 cache like you said, but that one would create a silly segmentation.
but hey it's amd, they love to offer 10000 cpus to cover each and every price point


Edit: just saw 3950x got postponed to November 😂
Seems they dont have enough (good) chips yet.

So it will be Epyc first, n then TR4>3950x probably as the latter two are scheduled for the same month.

This post has been edited by nrw: Sep 21 2019, 01:37 AM
nrw
post Sep 21 2019, 05:10 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(chocobo7779 @ Sep 21 2019, 12:01 PM)
Most probably the fact that TSMC is having some supply issues sweat.gif
*
nah, https://www.crn.com/news/components-periphe...pyc-rome-demand
nrw
post Sep 24 2019, 04:36 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


quite funny, with b550 coming within the next 5 weeks (oem's are listing it for 2nd week of October) the msi b450 max series probably gonna have a pretty short lifetime.
confirmed they wanted to use the hype to clear stock just exchanging the bios chip.

at least prices gonna plummet. Or well, knowing Malaysian retailers I rather say.. drop marginally.

This post has been edited by nrw: Sep 24 2019, 04:38 PM
nrw
post Sep 24 2019, 06:10 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(vernon91 @ Sep 24 2019, 05:04 PM)
the thing is we don even know what kind of feature can the b550 chipset offer over b450 , since the x570 major different is PCIE gen4 hmm.gif

hopefully is not a re branding of the b450 to b550 sweat.gif
*
huh? the specs are released.

I attach a German site's overview, cuz that's where I read about it yesterday.
Actually this chipset was meant to be released alongside X570.
But as ASMedia wasn't capable of delivering in time it's being released 3 months late.

It's quite obviously is a successor to b450 which is even more capable than x470 boards.
Hence B450 and X470 will drop in price quite a bit as those B550 chipsets are ~$30USD cheaper than those sold to board manufacturers by AMD for X570.

The only question left open is whether or not usb 3.0 will be using pci lanes, or not. Other than that everything is specified quite clearly.

This post has been edited by nrw: Sep 24 2019, 06:13 PM
nrw
post Sep 25 2019, 03:49 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(Bonchi @ Sep 25 2019, 03:35 PM)
my problem is the spikes aka boost clocks plus high voltages tends to be at 70% of the time. It caught my attention because my fan kept spinning at max speed even from just web browsing with music player playing at the background without monitoring tools... seems like the processor is going at 60C+ with such minor tasks because that's where i set my fan to go 90%. (These slim 80mm delta high RPM fans are no joke in terms of loudness too sweat.gif)

The idle and cores going to sleep only seem to happen when i turn off everything and stare at the desktop wallpaper.

I've got no choice put to set the processor max speed to 99% in power management. meaning it will never boost and that drops the temp by 10-15C  but i lose out on the performance spec from the boost. This is really tough..... plus i dont think this is the right behavior for the core clock mapping.
*
if you can't manage to solve it with fancurves and do not want to 1) accept the noise or 2) change your cooler you could always clock it at a lower voltage. That would be a solution between default boost and no boost.
which cpu u got? 3600?

This post has been edited by nrw: Sep 25 2019, 03:49 PM
nrw
post Sep 25 2019, 04:17 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(Bonchi @ Sep 25 2019, 04:09 PM)
yup 3600 luckily i didnt go for 3700x knowing the temps on the 3600. cooler is NHL9A which is the only cooler that fits in my super tiny ITX case. I guess i'll just have to live with it without boost... 

Well it's not like as if I can feel that it's significantly slower for daily use without the boost.... Just that it's a little bit irritating that you're not utilizing its full potential. Sommore the max load is 75C only. The only problem is with web browsing temps tend to remain at 60 ish because it's boosting. Meanwhile intel would stay at 1+ Ghz only.

Hopefully amd will one day fix their clock stepping based on CPU utilization a little better
*
how about using ryzen master then?

Try 1.20V with 4000MHz. Maybe that will drop temperatures in a region that benefits your situation.
Still better than being stuck at 3.6,

Alternatively you could just set your noctua to spin at a fixed rpm which is acceptable to you?
nrw
post Sep 25 2019, 04:38 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(krayden @ Sep 25 2019, 04:22 PM)
Need help with cross hair hero 8 bios update

3800x
16gb trident z neo
CH8

Flashback failed (steady blue light, usb2)
Ez something in bios update, (file not bios), over internet (stuck at 8%)
Tried rename bios file to CH8.Cap and without renaming.
Ez update in windows, managed to get it updated to 802. The program doesn't see 902.

Any ideas?

BTW, I dunno if I'm reading it wrong, but in hwinfo all cores max out at 4.397 considered not too bad right? Advertised boost at 4.5.
*
flashing from windows... naaay.
I only have the VII Hero for a short time but flashback should be pretty fail proof.

Are you sure you're using the correct verison (non wifi vs wifi version) and naming the file accordingly?
I see your renaming attempt there and without checking I'm pretty sure there's two errors there.
Firstly I think for the C7H the filename was c7h.cap or c7hwifi.cap and secondly I only saw the WiFi version up for sale in Malaysia which should have another name.
So it should be C8Hwifi.cap

This post has been edited by nrw: Sep 25 2019, 04:39 PM
nrw
post Sep 26 2019, 09:54 AM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(lavenzo @ Sep 26 2019, 08:16 AM)
hi, my gigabyte x570 aorus pro wifi arrived and the mobo came with bios version F2, website listed F3 is the initial version. is my board pre-production beta testing board??

i then flash f6b via qflash but system having weird behavior, i have trident z neo 3600 ram and XMP enabled in bios already but go in windows cpuz showing 2133 only and back bios check and XMP still enabled

then my cinibench r20 is quite low too, im getting only like 4300+, im using 3700x.

is mobo issue or ram issue?
*
cinebench doesn't bother about ram too much when it comes to scoring. but 4300 is way too low.
chipset driver,... all installed?

for the ramspeed it could be that the xmp doesn't stick, just key the values manually and do not forget to raise the fclk accordingly (1800 for your 3600 ram setting)
nrw
post Sep 26 2019, 12:00 PM

Casual
***
Junior Member
398 posts

Joined: Jul 2014


QUOTE(lavenzo @ Sep 26 2019, 11:48 AM)
i not sure if latest driver installed. i installed a free driver sw and just run to update all driver. is there another way to get these chipset updated?

i notice another thing, if i stick ram to A2 & B2 then XMP works but not A1 & B1..why ah? mobo issue?
*
maybe next time you take a look into the manual then. pretty sure the default config for 2 ram sticks will be a2+b2...

furthermore I'm sure chipset drivers are also offered on the support section for your mb, alternatively you can just dl from amd.

https://www.amd.com/en/support/chipsets/amd-socket-am4/x570

17 Pages « < 2 3 4 5 6 > » Top
 

Change to:
| Lo-Fi Version
0.2118sec    0.06    7 queries    GZIP Disabled
Time is now: 7th December 2025 - 03:32 PM