Who's Happy So Far with Ryzen 3000 Series?

Yes they should. Did you change any of the sub timings? If possible compare your current sub timings vs the 3600 c16. It may be that one or two of them are set too loose.
 
is your infinity fabric running at 1800mhz? It should be running at ram frequency for the best latency behavior, which would be 1800 for 3600ddr.

It could be that with infinity fabric hitting 1800mhz, you are hitting your tdp more easily and so not boosting as high or as often as other tests with lower ram speed. PBO may be needed to push tdp limits up ....though Asus boards and pbo seem to be not on speaking terms for most of their lineup. Behavior of enabling pbo is leaving much to be desired.
 
20190825_173557.jpg
Yes they should. Did you change any of the sub timings? If possible compare your current sub timings vs the 3600 c16. It may be that one or two of them are set too loose.

I got my subtimings from the Ryzen DRAM Calculator Safe Preset for 3600 with B-Dies, 2 Rank, 2 DIMMS, V1, X570. My Windows install is kind of screwed. I've been waiting to get everything stable before re-installing. I wonder if that would help. Then again, it's been screwed for a while and I've gotten faster results recently with looser settings.
 
Last edited:
is your infinity fabric running at 1800mhz? It should be running at ram frequency for the best latency behavior, which would be 1800 for 3600ddr.

It could be that with infinity fabric hitting 1800mhz, you are hitting your tdp more easily and so not boosting as high or as often as other tests with lower ram speed. PBO may be needed to push tdp limits up ....though Asus boards and pbo seem to be not on speaking terms for most of their lineup. Behavior of enabling pbo is leaving much to be desired.

Yeah, Infinity Fabric is at 1800 along with the RAM. My BIOS does seem to revert settings sometimes on reboot when I have to hit the retry button, specifically, I noticed once that it had taken all the new RAM timings, but fclk was 1866 and memory clock was still at 3733 instead of 1800 and 3600. I wonder if fclk was at 1866 during that run. Doesn't seem likely though. I may reload my BIOS profile and try again to see what happens.
 
I did. The graphs show the opposite of what was claimed.

The response and graph was directed toward someone playing CSGO. The graph shows the AMD processor as faster in CSGO (ever so slightly). Other games which were not referenced by the poster obviously show the 9900k in front of AMD parts by various margins (I'm not here to debate the significance of either ;) ). I'm not sure what you're referencing...
 
The response and graph was directed toward someone playing CSGO. The graph shows the AMD processor as faster in CSGO (ever so slightly). Other games which were not referenced by the poster obviously show the 9900k in front of AMD parts by various margins (I'm not here to debate the significance of either ;) ). I'm not sure what you're referencing...

That the description of 'slightly' that you use is an overstatement- it's a margin of error- and that the description by the poster making the argument was so overstated as to be comical.
 
The response and graph was directed toward someone playing CSGO. The graph shows the AMD processor as faster in CSGO (ever so slightly). Other games which were not referenced by the poster obviously show the 9900k in front of AMD parts by various margins (I'm not here to debate the significance of either ;) ). I'm not sure what you're referencing...

He was wrong but he wont admit it. It was clearly stated it was CSGO and now wants to try to flip it so he doesn't look bad. Way too many assume a 5.0GHz all core overclocked 9900K is a automatic win in gaming and that is not always the case, most times it's still within the margin of error or around 10 fps or less and will have 0 effect on your gaming experience. Both work great for gaming and one is a ton better at certain workloads.
 
He was wrong but he wont admit it.

Lol.

It was clearly stated it was CSGO and now wants to try to flip it so he doesn't look bad. Way too many assume a 5.0GHz all core overclocked 9900K is a automatic win in gaming and that is not always the case, most times it's still within the margin of error or around 10 fps or less and will have 0 effect on your gaming experience.

Using CSGO here isn't a metric of performance- it's a test to make sure that there's nothing wrong. Since there isn't, it's an extreme margin of error.

Both work great for gaming and one is a ton better at certain workloads.

And you imply that I'm claiming otherwise, which is ridiculous. I dare you to quote me in support of your falsity.
 
That the description of 'slightly' that you use is an overstatement- it's a margin of error- and that the description by the poster making the argument was so overstated as to be comical.

But the response and the graph, etc. was to a poster claiming that CSGO suffered using an AMD processor compared to an Intel one, which clearly isn't the case because, as you say, it's a margin of error.

Wish the games I play used the AMD more effectively. I only play CSGO and DCS Combat Sim. Priority is frequency in these games and my Intel 8086k@5ghz is faster :(. Once AMD can match Intel in my games I'll jump ship, but I'm not going to spend $ on an "upgrade" and lose speed.

Here's the quote. Obviously, there isn't much point in moving off of an 8086k to upgrade to anything if you're just playing CSGO.
 
Lol.



Using CSGO here isn't a metric of performance- it's a test to make sure that there's nothing wrong. Since there isn't, it's an extreme margin of error.



And you imply that I'm claiming otherwise, which is ridiculous. I dare you to quote me in support of your falsity.


Making friends again ? :eek::eek::eek:



(I'm jk)
 
I don't react well to implications of bias from those with very obvious bias themselves ;)

Your in a owners thread without owning a Ryzen 3000 series chip and trying to argue about things and were wrong on your assumption and now suddenly it's within the margin of error when that was not what you implied. Not only that but you decided to chime in on someone that was correcting a very false claim that Ryzen was slower on CSGO and then now are trying to change the story of what you meant by that post. People with a obvious bias never admit they are wrong they just try to obfuscate the argument until no one knows what the original disagreement was in the first place. Heck even Kyle has yelled at you for doing that.

Benchmark proved your statement wrong as Ryzen is faster by a tiny margin over the overclocked 9900K, which tends to highlight how little difference there is between the two in most benchmarks. It might also be why they are flying off the shelves as people are starting to notice the difference in gaming is pretty tiny, while productivity is better on Ryzen. As we can see in this thread most people have been happy with their chip.
https://www.amazon.com/Best-Sellers-Computers-Accessories-Computer-CPU-Processors/zgbs/pc/229189
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
I posted in the Zen 2 Memory Speeds thread this morning. Not really happy yet. I'm trying to get RAM B Dies rated for (4000 19-19-19-39) stable at 3733 with decent timings and low latency without much luck so far. I want to get everything worked out, then reinstall a fresh copy of Windows. One crash I had yesterday broke Windows networking pretty hard. I had to use some google fu to fix it. 3900x on an Asus Crosshair Hero VIII. I'm sure once I get stable, I'll be happy. Maybe another new BIOS is needed.
Your better off sticking with tighter timings and lower ram speed. Are you decoupling the FI speed from the ram. 1:1 does give you the best but it gets flaky above 1800MHZ or DDR4 3600 speeds. Also going above 3600 on mem speeds have shown to affect PBO and core boost speeds.
 
20190827_230332.jpg
Your better off sticking with tighter timings and lower ram speed. Are you decoupling the FI speed from the ram. 1:1 does give you the best but it gets flaky above 1800MHZ or DDR4 3600 speeds. Also going above 3600 on mem speeds have shown to affect PBO and core boost speeds.

Well, last night I got a successful memtest86 pass at 3733 mem, 1866 fclk 16-16-16-32. Thanks to Darth and Cciti recommendations. I think the key was running SOC at 1.125 and cranking more of the settings for VDDG, VPp, VTT, etc. Latency is now 65.1 ns in Aida. I used RAM timings from the DRAM calculator Fast setting for 3733.

You may be on to something about core boost speeds vs. RAM speed though. I'm scoring below 7200 in Cinebench R20 multicore.
 
Last edited:
Parts incoming!

Asus Crosshair VIII Hero
3700x
32GB Team Group / 8Pack 3600Mhz CL16
Corsair MP600 1TB
Dark Rock Pro 4

This will replace the 4770k in my sig. 1080Ti SLI will transfer over. Hopefully, this will also last as long as the 4770k did!
 
Parts incoming!

Asus Crosshair VIII Hero
3700x
32GB Team Group / 8Pack 3600Mhz CL16
Corsair MP600 1TB
Dark Rock Pro 4

This will replace the 4770k in my sig. 1080Ti SLI will transfer over. Hopefully, this will also last as long as the 4770k did!

I have a similar setup. Works great. Good luck with the build.
 
Right, my system is built, C8H, 3700x etc. (as above). Time Spy has gone from 12.5k to 17.5k. Cinebench R20 coming out at 4636 after 10 passes.

Question about clocks and temps. I'm currently running prime95, it's been going for maybe 10 mins. According to Ryzen Master, my clocks are 3.75GHz across all cores. CPU voltage is 1.1V and temps are 63.5*C give or take - doesn't seem to get any hotter than this.

Is that right? Should it be boosting higher?
 
Running Prime 95 I get all cores at 4.1 Ghz and 63 degrees at 1.36 volts. No overclock just the 3900x running with a Corsair H60 AIO, normally idles at 4.25Ghz. Happy so far, wondering if I should upgrade to a x570 board, have a B450 now running 90% on 168A EDC according to Ryzen Master. Not sure if that maters, been a while since I built a system and cared to make it reach its max.
 
Running Prime 95 I get all cores at 4.1 Ghz and 63 degrees at 1.36 volts. No overclock just the 3900x running with a Corsair H60 AIO, normally idles at 4.25Ghz. Happy so far, wondering if I should upgrade to a x570 board, have a B450 now running 90% on 168A EDC according to Ryzen Master. Not sure if that maters, been a while since I built a system and cared to make it reach its max.

I'm think you try better cooling and see if there's any change with your setup as is.
 
I'm think you try better cooling and see if there's any change with your setup as is.

That is the third cooling option I have tried. Had a Corsair H100 AIO and the Cooler that came with the processor. When I clocked it in the Bios to 4.3 it was fine, clocked it to 4.4 and it would not even boot, just looped. Temps are actually lower with the new cooler. Can't fit a triple fan AIO in the case as the MSI boards upper hear sink and RAM are in the way of the fans. I was looking at a Noctua D15 or D15S as well as a BeQuiet Dark Rock Pro but can not tell if they would clear the RAM, I have 4 Corsair Vengence Pro RGB 3600 chips right there that they both look like they would cover but the chips are not exactly short. Or else I would buy one of them. I am also curious about some of those power levels and if the processor needs more power that the Board can not supply or is at its limit. Last time I was Overclocking anything it was pretty basic compared to now. And from what I have seen so far of course hitting that 4.6Ghz number is just a boost, I am running 4.25Ghz all day long with no issues and the processor sitting around 40-50 degrees.
 
Right, my system is built, C8H, 3700x etc. (as above). Time Spy has gone from 12.5k to 17.5k. Cinebench R20 coming out at 4636 after 10 passes.

Question about clocks and temps. I'm currently running prime95, it's been going for maybe 10 mins. According to Ryzen Master, my clocks are 3.75GHz across all cores. CPU voltage is 1.1V and temps are 63.5*C give or take - doesn't seem to get any hotter than this.

Is that right? Should it be boosting higher?

check hwinfo, those numbers look all wrong.. 1.1v is probably the SOC voltage, not cpu voltage. if hwinfo says it's 1.1v you may need to check the bios because somethings set wrong.
 
That is the third cooling option I have tried. Had a Corsair H100 AIO and the Cooler that came with the processor. When I clocked it in the Bios to 4.3 it was fine, clocked it to 4.4 and it would not even boot, just looped. Temps are actually lower with the new cooler. Can't fit a triple fan AIO in the case as the MSI boards upper hear sink and RAM are in the way of the fans. I was looking at a Noctua D15 or D15S as well as a BeQuiet Dark Rock Pro but can not tell if they would clear the RAM, I have 4 Corsair Vengence Pro RGB 3600 chips right there that they both look like they would cover but the chips are not exactly short. Or else I would buy one of them. I am also curious about some of those power levels and if the processor needs more power that the Board can not supply or is at its limit. Last time I was Overclocking anything it was pretty basic compared to now. And from what I have seen so far of course hitting that 4.6Ghz number is just a boost, I am running 4.25Ghz all day long with no issues and the processor sitting around 40-50 degrees.


I was thinking more along the lines of grabbing a $$$ EK prefab open loop kit off Amazon then returning it when you are done.

Check that box off, bc a high end air cooler or aio is only going to get you so much.

Open loop just give you more thermal capacity, or time before heat saturation.

It'd just answer the question of what your particular CPU is capable of doing.

I posted the derbauer clip, take a look at his survey for what normal reported boost range for your cpu is.
 
Pretty damning. It appears that it's a silicon lottery just to hit advertised specs.

That was why I posted the clip.
If I was some of these posters 400mhz under advertised box spec I'd want to know if my mobo, CPU, or cooling were holding back my build.

Otherwise I want $50 back from AMD.

I'm on a cheapie 2600 build, so it's not like I'm going to wrangle every drop of what I've got into it and get annoyed if it fell short.

If I'm $450+ into my CPU then yes, I want to be as close to 4.6 as possible.

If I *had* to piece together an open loop after trying a prefab kit, it's easy and cheap enough to do.

X570 can be had cheap enough if the B450 Mobo is somehow holding him back.....I kinda doubt it.

I'd want to try a different 3900x and see if there was any difference.
 
I was thinking more along the lines of grabbing a $$$ EK prefab open loop kit off Amazon then returning it when you are done.

Check that box off, bc a high end air cooler or aio is only going to get you so much.

Open loop just give you more thermal capacity, or time before heat saturation.

It'd just answer the question of what your particular CPU is capable of doing.

I posted the derbauer clip, take a look at his survey for what normal reported boost range for your cpu is.


As fun as that would be I ordered the Noctua D15S and before installing will see if it clears my DIMMS. Looks like it should with the single fan. I think the 4.25 is all I will get without manual overclocking it. Which I can get a solid 4.3Ghz all 12 cores, although I did not run Prime 95 on that when I did have it set that way. It is not like it is some slouch of a processor for what I do anyway, just figured I would try to see what its max is. I mean in the perfect world I would clock it to all core at 4.6 and let it heat the room for me. Like a space heater. I just wonder if I need the 570 to get the most or if my 450 will do.
 
As fun as that would be I ordered the Noctua D15S and before installing will see if it clears my DIMMS. Looks like it should with the single fan. I think the 4.25 is all I will get without manual overclocking it. Which I can get a solid 4.3Ghz all 12 cores, although I did not run Prime 95 on that when I did have it set that way. It is not like it is some slouch of a processor for what I do anyway, just figured I would try to see what its max is. I mean in the perfect world I would clock it to all core at 4.6 and let it heat the room for me. Like a space heater. I just wonder if I need the 570 to get the most or if my 450 will do.

Hwinfo and see if anything on your Mobo is topping out voltage or heat.

You can test for core throttling.

Simple as that.
 
Hwinfo and see if anything on your Mobo is topping out voltage or heat.

You can test for core throttling.

Simple as that.

Only thing I see coming close to its limits is EDC according to Ryzen Master they are over 90% of 168A. After 10 Minutes on Prime 95 TDC goes to 100% and the cores drop from 4.1 to 3.85 and temps hits 90 degrees. Before that it sits at around 65-70. Powers look good. MOS temps go up to 90 degtrees though at this point from 60 before. The picture is after 15 minutes all cores default test that opens with Prime 95.
 

Attachments

  • Prime95.jpg
    Prime95.jpg
    269.5 KB · Views: 0
  • Prime95-2.jpg
    Prime95-2.jpg
    245.4 KB · Views: 0
Only thing I see coming close to its limits is EDC according to Ryzen Master they are over 90% of 168A. After 10 Minutes on Prime 95 TDC goes to 100% and the cores drop from 4.1 to 3.85 and temps hits 90 degrees. Before that it sits at around 65-70. Powers look good. MOS temps go up to 90 degtrees though at this point from 60 before. The picture is after 15 minutes all cores default test that opens with Prime 95.


I'm thinking open loop would be a good place to explore.

I'm thinking there's some more there.
 
seeing prime95 cause the cores to drop to < 4Ghz is not uncommon when doing the most stressful test (and avx is enabled). It's not a temperature thing. It's a tdp thing. I have a 3900x and a 3800x. Both on different asus x570 motherboards. With prime95 doing it's worst, the cpu becomes heavily tdp restricted.

There are certain load types that allow significantly exceeding the tdp. It's just not via the kind of full cpu load prime95 can generate.


And of course, if you can't keep it cool, temperature will become a factor if you're seeing temps in the 90C range. But even if you kept it in the 80's, you'd see the same kind of significant drop in frequency.

At least from what i've seen and tested.
 
Upgraded from a 2700x x370 to a 3800x and x570. Quicker boot and load times. Crunching FAH work units like crazy at a steady 4.1gh. Solid processor.:)
 
AMD is releasing a new bios fix for precision boost coming in the next week to various motherboard manufacturers.

Should be interesting to see how this impacts performance. Especially on the 3800 and 3900.


Now if we could just get motherboard manufacturers to use ACPI tables correctly ...
 
  • Like
Reactions: mjz_5
like this
seeing prime95 cause the cores to drop to < 4Ghz is not uncommon when doing the most stressful test (and avx is enabled). It's not a temperature thing. It's a tdp thing. I have a 3900x and a 3800x. Both on different asus x570 motherboards. With prime95 doing it's worst, the cpu becomes heavily tdp restricted.

There are certain load types that allow significantly exceeding the tdp. It's just not via the kind of full cpu load prime95 can generate.


And of course, if you can't keep it cool, temperature will become a factor if you're seeing temps in the 90C range. But even if you kept it in the 80's, you'd see the same kind of significant drop in frequency.

At least from what i've seen and tested.

Yeah as you can see on the second picture the cores sped back up and temp dropped back down. They went back to the 4.1 area and stayed there. That was around the 20 minute mark of running it. I was just curious to know if there is any advantage to running a 570 board in my case over the 450? Or just stay with it since the board is only like 6 months old.
 
pcie 4 is the only reason why you would really benefit from x570. Also, I dont know if any 450 boards support 128GB of ram, which is also an option on some x570 boards.

it's unlikely you'll really tax the power system since most decent am4 boards since ryzen 1 should be able to handle the current draw and power requirements of even a 3900x.

While the ram controller is on the cpu, and so motherboards shouldn't matter when it comes to things like compatibility, it's obvious that there is a factor in quality of traces and other electrical aspects that are still heavily motherboard-centric and a newer x570 that is designed to handle ddr 3800+ will better work with any ram stick you throw at it, vs older boards that were not designed electrically to support such memory, and as such, may or may not be within tolerance for them.

You almost definitely wouldn't notice a 300 dollar difference in performance. I'd recommend staying with the 450 since you just bought it unless you have hardware that can actually make use of pcie4 or if your current ram is not really working at the speed it's supposed to.
 
pcie 4 is the only reason why you would really benefit from x570. Also, I dont know if any 450 boards support 128GB of ram, which is also an option on some x570 boards.

it's unlikely you'll really tax the power system since most decent am4 boards since ryzen 1 should be able to handle the current draw and power requirements of even a 3900x.

While the ram controller is on the cpu, and so motherboards shouldn't matter when it comes to things like compatibility, it's obvious that there is a factor in quality of traces and other electrical aspects that are still heavily motherboard-centric and a newer x570 that is designed to handle ddr 3800+ will better work with any ram stick you throw at it, vs older boards that were not designed electrically to support such memory, and as such, may or may not be within tolerance for them.

You almost definitely wouldn't notice a 300 dollar difference in performance. I'd recommend staying with the 450 since you just bought it unless you have hardware that can actually make use of pcie4 or if your current ram is not really working at the speed it's supposed to.

Well I do have DDR 3600 running at 4000 right now. :)

I currently have the MSI B450 Tomahawk. Was thinking to up to the MPG X570 Gaming Plus. The bonus would be then I would have most of the components for a whole other system, Ryzen 2600, MSI board, 32GB DDR 3000 memory, 650W PSU I just replaced and two different AIO coolers. Lol! All I need is a case and video card.
 
Back
Top