Welcome Guest ( Log In | Register )

22 Pages « < 7 8 9 10 11 > » Bottom

Outline · [ Standard ] · Linear+

 AMD Ryzen, AM4 / AM5 Platform

views
     
Bonchi
post Jan 23 2021, 12:40 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(JohnLai @ Jan 22 2021, 09:17 PM)
AGESA 1.2.0.0 doesn't seem to make any difference to my Infinity Fabric OC, still limited to 1866Mhz.  sad.gif
Speaking of which, which batch is your 5900x?
My 5950x batch is 2046 = year 2020 week 46.
*
it’s only guaranteed to 1600Mhz and the max is 2000. So anything more than 1600Mhz is see your luck. In fact before all the bios updates, many couldnt even run stable at 1600.

QUOTE(mystvearn @ Jan 23 2021, 02:15 AM)
Not x variant. Not OC for now. I plan to buy higher cpu few years later, used or whenever the price is reasonable.
*
AM4 will be discontinued this gen sweat.gif. Zen4 will require a new socket already.

This post has been edited by Bonchi: Jan 23 2021, 12:42 PM
Bonchi
post Jan 23 2021, 03:40 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(Jedi @ Jan 23 2021, 03:23 PM)
Every time I drop by this ryzen thread sure got some performance haggle issues reported (I havent even drop in OCN even)

It seems that this ryzen zen 3 is only good on paper, and also a straw launch. (not even paper launch, at least RTX 3080 and 3090s comes every 2 weeks or so)
Without a minimum of DDR4 3200 CL14 or DDR4 3600 CL16, I dont know if 5900x can beat i9 10900K. If theres something to learn, its that AMD are shills. And youtubers words cannot pakai wan if you buy a PC to game more than anything else - they run it on High end motherboards, under a controlled setting, maybe also paid reviews some of them.

U are not going to run cinebench, prime 95 all day long, certainly 80% gamers are casual group, also do not overclock like those youtubers do.
U buy the PC to instruct it to game lar. And instruct the CPU, I want you sit at this X Ghz all day long because this is the money I invested on you. 
Until now many of u are still running benchmarks to see whether the zen 3 stable ka, temperature ok ka, voltage ok ka. 2 months + already, waiting for AGESA updates (stability stability stability every update is it even a stability update why every time call it stability update)

Finally good luck playing the bsod roulette.
*
AMD did one thing great... taking the price of intel down rclxms.gif
Bonchi
post Jan 23 2021, 04:30 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(Cold|Drawn @ Jan 23 2021, 03:55 PM)
depends on how you view things, some of us just like to play around to squeeze more performance out of it.
just like intel, if you OC, you'll definitely run into instability. It's part of the deal.

sometimes, we post here just to see if there's anyone with similar experience so others might share their experience to aid others.

Same goes to intel even in OCN, people share their experience or difficulty and see if there are pros to help them OC better. Bet that's what CST does in OCN to help others too.

Problem is when people see this sort of posts then witch hunt it like some sort of plague.
*
but problem is we're not talking about OC or tweakers.. Some ryzen chips is giving issues at stock not even overclocked. Sometimes even with ram at 2444mhz despite stating "natively support" 3200mhz. Requires a bunch bios updates, waiting version after version... until by the time when a new proc is released, some of the issues still not entirely solved.
Bonchi
post Jan 23 2021, 06:35 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(Cold|Drawn @ Jan 23 2021, 05:50 PM)
it's ok to state a problem, AMD users would understand tbh. I've had issues as well
and this exact thread encourages people to post their problems too.

but too many somes only tell one thing, bonchi.
Many instances that you pointed out are probably past examples that you pulled off your memory
could have been resolved like ya said with bios? but you still held on to it.

then amplify the issue by trying to chain another instance and stack on to the past.
hmm, I'm just questioning how true are the claims?
*
it’s not past, it’s still happening in fact again just few days ago as you can see in my postings with latest bios. It still crashes occasionally despite already passed more than hundred of hours of stress test. And now the bios support for zen2 has long been stopped looking at the few latest changelogs. We can say AMD is giving zero crap about zen2 already.

For now I took ctskl advice and try tweak the LLC to high and i have PBO disabled all this while too. And im still monitoring to see if it will crash... Which is totally ridiculous because im just running at stock. And it doesnt crash frequently with any known trigger which makes it even more annoying. It can run stable for more than a month then suddenly crash again when youre in the middle of typing something on words or in a game’s mission.

If you think all the claims are false.. Why would I wanna bash and exaggerate on my own hardware which i have spent so much money on. In fact i have swapped every single component of my pc already even the casing and monitor, mouse and keyboard too and still have this problem.

This post has been edited by Bonchi: Jan 23 2021, 06:38 PM
Bonchi
post Jan 23 2021, 07:24 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(yimingwuzere @ Jan 23 2021, 06:56 PM)
TBH 99% of people who post comments like yours usually are fanboys blowing air on their products and actually don't know anyone behind the scenes. If your sources are right, you can link back to that and say "told you so" once the news actually comes out lol.

DDR4-6000 on air sounds great nonetheless. revE or B-die though?

At any rate that poster on 4650G and the RAM LN2 pot shows one huge advantage monolithic has over chiplet design - RAM speeds can go higher on Zen2/3 APUs vs the Zen2/3 7nm + 12nm IO die design. I'm struggling to get my 5800X + Asrock B550 Phantom Gaming ITX + Ballistix 3200C16 dual rank to 3800 in coupled mode. The RAM does 3800 easily when decoupled from IF, I can boot IF on 3800 and pass TM5 anta777 config with my RAM kit on XMP timings too, but can't seem to do both together. Not sure if it's down to CPU, or BIOS at this point.
*
the whole point of fast ram being beneficial on ryzen is for the coupled mode to increase the IF bandwidth. Without that it’s quite pointless for reaching those ram speeds but sacrificing on the interconnect bandwidth.

If there are other people claiming they can reach those speeds with the same bios then it’s probably down to the cpu/ram/psu already.


Bonchi
post Jan 23 2021, 07:27 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(Cold|Drawn @ Jan 23 2021, 07:21 PM)
no worries, I get the frustration when you won't know when WHEA creeps onto you.
My CP2077 gameplay crashed left/right until I was forced to grind so I can end a mission quick before the game crash.
To the extent that I reset bios stock, barebone stock ram, cpu, gpu all stock but then still crashed.
Friend's intel plays the game just fine with no crashes.

ever considered you got a potato chip?
*
i sold my previous chip and got a new one of much later revision. still the same.

As i mentioned it happens quite rarely so probably most users do have this problem as well but they just shrug it off, especially those who runs overclock daily. So no one really come up to highlight this i guess.
Bonchi
post Jan 23 2021, 08:45 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(cstkl1 @ Jan 23 2021, 08:21 PM)
they were already winning.

they could have launched a monolithic version of 5600/5800x and won the gaming part.

so far i see 5950x user the most hit. literally 50:50 can work.
*
the whole reason they didnt use monolithic die is just so they can increase their core yields and thus able to reach so many core counts... plus their IOD is 12nm.

QUOTE(SSJBen @ Jan 23 2021, 08:40 PM)
Single CCD has issues with heat. Rarely see 5800x user not complain about temps.
*
single CCD will also occasionally have weak cores which is a ticking time bomb, like what im facing (with 2 diff CPUs) Hopefully a more aggressive LLC will solve the issue at the cost of running on higher voltages and temps.

This post has been edited by Bonchi: Jan 23 2021, 08:51 PM
Bonchi
post Jan 23 2021, 11:50 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(yimingwuzere @ Jan 23 2021, 11:46 PM)
It's just an epeen contest with high RAM speeds. Although IIRC Anandtech did review a Corsair 5000 kit that showed a small improvement over the fastest kits for Ryzen in spite of the decoupled IF/uncore to RAM clock.
*
Yeah you need to get until that level to see improvement. meanwhile staying at 3800 or even 3600 might be better overall.
Bonchi
post Jan 24 2021, 11:03 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(SSJBen @ Jan 24 2021, 10:17 PM)
There is also another aspect to be put into perspective.

A lot of Zen 2 and 3 owners, many of them were upgrading from Sandy Bridge/Ivy Bridge era. So the perf jump that they are experiencing has already skewed their perception into how large of a jump it has been, thus creating the AMD love.
*
I jumped from an Ivybridge 4770K system and from a broadwell macbook pro and the ryzen's temp gave me a shock laugh.gif ... and the performance difference is negligible considering the old intels are running on DDR3.

I actually spent months solving the average use(such as browsing) temp issues that goes to 50-60C, going through a few coolers, bios, fan mods, case mods and settings to ultimately give up altogether... end up settling with a 120mm aio with a very silent fan spinning at a minimum of 1500RPM so the temp stays below 50C

This post has been edited by Bonchi: Jan 24 2021, 11:06 PM
Bonchi
post Jan 25 2021, 12:12 AM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(yimingwuzere @ Jan 24 2021, 11:55 PM)
Maybe next step is the der8auer offset mount for AM4 so the AIO's centerpoint is closer to the CPU die instead of the center of the IHS? Assuming it fits your ITX board of course.
i5 quads to Ryzen will show a far bigger difference than i7 quads upgrading, assuming the i7s were decently clocked (4.4-4.6GHz). HT does make a difference nowadays, even if it didn't for gaming back in 2013.
*
next step is rocketlake. Had enough messing around with AMD and wouldn't bother investing anymore on this AM4 platform that is bound to be obsolete very soon... along with the Zen2 bugs AMD will never bother to fix.

Already gone through 2x 3600 CPU, 2x motherboards, 3x sets of rams, 3x psu... which I bought mostly new except 1 of the ram and bear all the depreciation costs to sell as used.

Plus it's not worth to spend so much especially on shipment for a signature custom mount just to try cool a 3600 which quite arguably wont make much of a difference on lighter semi idle loads, which is the main problem.... these are all buggy idle/load state issues on the Zen2 which seriously needs fixing as it may cause crashes but everyone seems to be shrugging it off saying it's normal... and AMD just moved on with zen3.

AMD did suggest using their ryzen’s balance powerplan that sets the minimum idle state to 99% may solve the crashing but having the cpu staying at 1.3V and touching 50C temps is not exactly a good solution either.


This post has been edited by Bonchi: Jan 25 2021, 01:23 AM
Bonchi
post Jan 25 2021, 02:15 AM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(SSJBen @ Jan 25 2021, 01:19 AM)
Agree. Hope Rocket Lake does well too so I have a reason to go back to Intel as well.
*
more like i hope rocketlake is cheap especially the B560 board. Really getting very tight on budget already... but still hoping to get off this platform. Even a 10400F or 10100F on a z490 ITX swap offer i will take without hesitation.

This “AMD experience” has blown my budget. it was originally meant to be a super low budget ITX build under RM4K which ends up literally causing me to buy parts enough to build 2 rigs and sell 1 to solve some issues... costing over RM2K in losses due to depreciation on selling all those used parts... And the random crashing issue is still not solved. bangwall.gif

This post has been edited by Bonchi: Jan 25 2021, 02:20 AM
Bonchi
post Jan 25 2021, 07:41 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(yimingwuzere @ Jan 25 2021, 07:24 PM)
Intel better not screw this one up again with value Core i5s.

8 series got away with it due to Zen/Zen+ arch being less competitive in gaming. Built a few 8400 rigs for friends with that, bypasses the need to overclock a 2600 + RAM all the way and end up consuming more power.

9 series i5s was pretty weak vs 3600.

10 series i5s at least added HT, even then B460 locking memory speeds to Intel's CPU max official support meant that 3600 and a decent 3200 kit still trades blows with 10400 and 2666 RAM while being better at non latency-sensitive workloads.

Is B560 going to be like B460 and still use Intel max official support, but Intel will allow DDR4-3200? Or is it full overclocking for frequency too? Can never be too sure with tech journos reporting about it saying it's the latter when RKL supports >2666 RAM officially.
*
B560 lately confirmed to be memory unlocked but cpu multiplier still locked. That’s why ctskl kept hinting... 11700 non K with a 3800mhz ram will be the best value option at pumping out fps

10400F is quite wasted in terms of potential due to memory lock on B460. It could perform very well with a z490 but the value ratio is off which is like getting an x570 for 3600.

This post has been edited by Bonchi: Jan 25 2021, 07:48 PM
Bonchi
post Jan 25 2021, 09:40 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(JohnLai @ Jan 25 2021, 09:31 PM)
Bonchi
Where does your 5600x made at?

Cause I just read :
https://www.reddit.com/r/Amd/comments/l4hr5...ntext=8&depth=9
Surely this can't be true, or perhaps?
*
mine’s a 3600 from china. Doubt it’s a hardware issue.. looks more like a loadline issue from agesa and the powersettings.

This post has been edited by Bonchi: Jan 25 2021, 09:44 PM
Bonchi
post Jan 25 2021, 10:04 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(JohnLai @ Jan 25 2021, 09:41 PM)
Eh? 3600? I guess that comment ain't true? grin.gif
*
well.... many of intel used to be assembled in malaysia and those CPUs all manage to test the test of time so i doubt it's the problem with manufacturing. It's more likely the wafer batches from TSMC that coincidently got sent to malaysia all have problems... so I wouldn't say because it's made in malaysia then it will fail.
Bonchi
post Jan 26 2021, 03:23 AM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(cstkl1 @ Jan 25 2021, 11:16 PM)
nrw says u can do stock
so try to set power limit etc all to actual stock 105w
*
i kinda able to narrow down the crashing of my amd and it seems to be a memory related issue which is related to my audio playback and USB DAC.

Finally for once an error managed to popup from my audio player right before the pc crashes and it stated something about bad pool caller which point to my audioplayer’s directory... which is no surprise because i buffer my playlist into my ram to reduce disk io along with asio as playback output which already been giving problems from the begining.

Guess it was a combination of ryzen’s shitty memory architecture and usb interface and the lack of legacy code support that is the cause of the random crashes... which kinda explains that there are no problems under long stress tests because i never turn on the music during the tests. Never had this issue when i was using intel since core2duo to a bootcamp win10 macbook pro retina... problem only started with this ryzen.

I guess at the meantime I will have to heavily compromise by refrain from using ASIO and decreasing the file buffer size to see if it still crashes.... and then it’s goodbye AMD on the next upgrade.

This post has been edited by Bonchi: Jan 26 2021, 03:31 AM
Bonchi
post Jan 26 2021, 12:10 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(faiz2036 @ Jan 26 2021, 11:26 AM)
How about 10400?  drool.gif
*
Wont be considered a big upgrade considering you need to change the mobo as well. That’s why im looking for a trade laugh.gif

QUOTE(TristanX @ Jan 26 2021, 11:37 AM)
I still remember Asus P8Z77-M can be found at RM339 and its overclockable! I miss those days with highest end chipset at affordable prices. Now both camps going nuts with the prices due to higher VRM. doh.gif
*
And on top of that, Im paying the ITX tax... where a B550 can match the price of a z490.
Bonchi
post Jan 26 2021, 02:14 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(SSJBen @ Jan 26 2021, 01:48 PM)
Interesting, does the issue still occur if you use a lower mem speed?

Also, did you try with WASAPI output as well?
*
yes even at 3200 xmp it crashed before.. wasapi tend to mess up my volume mixer so i didnt use it. Currently compromising by using direct sound to monitor.

I guess this is the escalation of the usb playback device stuttering issue that happens occasionally which i mentioned last time.. This might be the underlying issue that was causing the auido stutters.. which I also noticed the cpu usage fly up to 100% and everything becomes laggy in that brief moment.
Bonchi
post Jan 26 2021, 04:16 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(xxboxx @ Jan 26 2021, 03:59 PM)
I remember he is using pcie gen 3 riser. unless he have upgrade.
*
QUOTE(nrw @ Jan 26 2021, 03:24 PM)
Bonchi if u r on b550/x570 u can also set pcie of ur gpu to 3.0.
that can fix quite some issues you've been describing as well.
*
yeap, it's already running on PCIE gen 3 already.

have not tried running for weeks without a riser.. but currently using my third pcie riser, so I guess we can rule out the riser being the problem...with all the money wasted doh.gif

however changing mobo to B550 did solve the PCI-E lane issue... but the stuttering is still there when it goes into idle standby occasionally (disabling display off standby in power settings did solve the stuttering from happening... but that means i have to manually turn off the monitor sweat.gif and secondly USB suspend set to disabled ).... I actually got about 2 PM from users asking about this USB DAC playback stuttering as well and I've suggested this solution.

the main issue is that it happens once in a bluemoon, so it's very hard to catch and super annoying.... have to wait like weeks to really see if the solution works.

user posted image
Anyways, here's my ryzen master... wonder if there's anything out of the ordinary.

This post has been edited by Bonchi: Jan 26 2021, 04:38 PM
Bonchi
post Jan 26 2021, 07:33 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(JohnLai @ Jan 26 2021, 05:36 PM)
Hmm? VSOC at 1.1V? Hmmm........Bonchi, maybe you want to lower the VSOC voltage to 1.0V
https://www.tweaktownforum.com/forum/tech-s...7277#post977277
*
because of crashing so i up the vsoc to this as a troubleshoot just to monitor.... and also set the LLC to high under ctskl advice.

my symptom is not those, that guy definitely just whack a noob value at overclocking his ram that made the vsoc go so high and enters a boot loop. Gigabyte’s bios is indeed a little too complicated to noobs.

This post has been edited by Bonchi: Jan 26 2021, 07:38 PM
Bonchi
post Jan 26 2021, 07:52 PM

KittehPowah
******
Senior Member
1,649 posts

Joined: Sep 2008
QUOTE(yimingwuzere @ Jan 26 2021, 07:43 PM)
Wonder what RAM kit are you using. Can it go to 3733 or even 3800?
We need an Intel thread... maybe Intel gurus here like @cstkl can make one.
*
micron B die D9XPF. can hit 3800 cl18 but due to itx case with limited airflow over the ram i keep it at 1.35v and 3600 cl18 is the settings i settled with after 20hours of memtest86.

this ram can go very high at low voltages but the latency is challenging to set.

22 Pages « < 7 8 9 10 11 > » Top
 

Change to:
| Lo-Fi Version
0.1845sec    0.89    7 queries    GZIP Disabled
Time is now: 12th December 2025 - 01:09 PM