The Ryzen 5 3600 Discussion Thread

[humbug]

Been a while since I got my 5900X and I thought let me run a quick test to compare it with my trusty 3600. What a huge uplift in scores a year on

Keep in mind this was not run on a fresh install, so scores would be even a little higher potentially like was the case in the quoted post. Oh and not bothered OC'ing this beast yet either. All stock

https://valid.x86.fr/ut4n0b

 

[TNA]

https://valid.x86.fr/ut4n0b

Nice. Seems your OC is helping out with the single core score there

I could do a quick and dirty oc to beat it, but it would not count unless it is daily used clocks imo, which I assume yours is. I even have PBO off. May mess about overclocking at a future date when I fancy, but right now no appetite for it. With the 3600 I had not done overclocking for 6+ years so I was itching to do it when I upgraded.

Huge upgrade for you that. Should last a long time these CPU’s. Only way I see myself upgrading now is by getting the itch due to some nice tech coming out. Won’t be because I need it.

 

[humbug]

Nice. Seems your OC is helping out with the single core score there

I could do a quick and dirty oc to beat it, but it would not count unless it is daily used clocks imo, which I assume yours is. I even have PBO off. May mess about overclocking at a future date when I fancy, but right now no appetite for it. With the 3600 I had not done overclocking for 6+ years so I was itching to do it when I upgraded.

Huge upgrade for you that. Should last a long time these CPU’s. Only way I see myself upgrading now is by getting the itch due to some nice tech coming out. Won’t be because I need it.

You could and yes those are my daily clocks

Play with the negative curve optimiser in conjunction with a boost off set.

 

[TNA]

You could and yes those are my daily clocks

Play with the negative curve optimiser in conjunction with a boost off set.

Out of curiosity and it litterally takes a few minutes I did a negative curve optimiser of 5 across all cores and a bump of 25mhz and got this:

Maybe I will leave it like this for a while and see how I get on

 

[humbug]

Out of curiosity and it litterally takes a few minutes I did a negative curve optimiser of 5 across all cores and a bump of 25mhz and got this:

Maybe I will leave it like this for a while and see how I get on

I'm boosting to about 5.05Ghz ST.

Anything from 4.6Ghz to 5Ghz MT depending on what its doing.

The CPU will always try to go for the highest clocks it can to with in a boost limit, on the 5800X that is supposed to be 4.7Ghz "Maximum Boost Clock 4.7Ghz" i have a feeling this is a bit of a lie as it actually does that or near in Cinebench MT, ST its actually 4.85Ghz and from what i have seen they all do.
Yours is 4.8Ghz, actually 4.95Ghz, the 5950X is 4.9Ghz, actually 5.05Ghz.

The boost off-set limit is +200Mhz, on previous BIOS it was +500Mhz, don't know why they reduced it. With it i top out at 5.05Ghz, which is what i'm getting.

If you set +200Mhz you could probably reach 5.15Ghz provided its stable, if not you might like to set a plus curve optimiser for ST runs as you're never going to get to 90c with just one thread, for the highest MT boost is where your temperatures come in, anything over 60c your clocks start to drop from your highest boost, if i keep mine under 80c its around 4.7Ghz on all cores in Cinebech, with yours being a 12 core with the same 105 Watt TDP you might only get 4.4 to 4.5Ghz on all 12 in Cinebench., negative curve optimiser is where you keep your temperatures down, its like undervolting but its much more than that.

Its interesting how these SKU's are arranged.

5600X 65 Watts: 4.5Ghz all core
5800X 105 Watts: 4.7Ghz all core
5900X 105 Watts: 4.5Ghz all core
5950X 105 Watts: 4.3Ghz all core

Or something like that.

 

[TNA]

I'm boosting to about 5.05Ghz ST.

Anything from 4.6Ghz to 5Ghz MT depending on what its doing.

The CPU will always try to go for the highest clocks it can to with in a boost limit, on the 5800X that is supposed to be 4.7Ghz "Maximum Boost Clock 4.7Ghz" i have a feeling this is a bit of a lie as it actually does that or near in Cinebench MT, ST its actually 4.85Ghz and from what i have seen they all do.
Yours is 4.8Ghz, actually 4.95Ghz, the 5950X is 4.9Ghz, actually 5.05Ghz.

The boost off-set limit is +200Mhz, on previous BIOS it was +500Mhz, don't know why they reduced it. With it i top out at 5.05Ghz, which is what i'm getting.

If you set +200Mhz you could probably reach 5.15Ghz provided its stable, if not you might like to set a plus curve optimiser for ST runs as you're never going to get to 90c with just one thread, for the highest MT boost is where your temperatures come in, anything over 60c your clocks start to drop from your highest boost, if i keep mine under 80c its around 4.7Ghz on all cores in Cinebech, with yours being a 12 core with the same 105 Watt TDP you might only get 4.4 to 4.5Ghz on all 12 in Cinebench., negative curve optimiser is where you keep your temperatures down, its like undervolting but its much more than that.

Its interesting how these SKU's are arranged.

5600X 65 Watts: 4.5Ghz all core
5800X 105 Watts: 4.7Ghz all core
5900X 105 Watts: 4.5Ghz all core
5950X 105 Watts: 4.3Ghz all core

Or something like that.

Interesting. I had a look in HWINFO and I am just shy of 5GHz when it is running single thread and for multi it is around 4.45GHz.

Weird though how with 100Mhz less I am beating your score on single core.

I though I read somewhere LtMatt saying your off-set limit should not be more than 75MHz? I may up mine from 25 to 50MHz and do another test now

 

[humbug]

Interesting. I had a look in HWINFO and I am just shy of 5GHz when it is running single thread and for multi it is around 4.45GHz.

Weird though how with 100Mhz less I am beating your score on single core.

I though I read somewhere LtMatt saying your off-set limit should not be more than 75MHz? I may up mine from 25 to 50MHz and do another test now

4.975Ghz with the 25Mhz off-set, which is 75Mhz less, a 1.5% difference and your score is basically identical.

Margin of error, maybe you have better memory, less things running in the background when you run it..........

 

[TNA]

So did the test, not much in it to be honest. I just turned it all off as the performance difference is so little. I think to get the most out of it I would have to mess about finding what is best offset for each core and as I say right now can't be asked. Had a little fun just now though

 

[humbug]

So did the test, not much in it to be honest. I just turned it all off as the performance difference is so little. I think to get the most out of it I would have to mess about finding what is best offset for each core and as I say right now can't be asked. Had a little fun just now though

I know the feeling, some cores are better than others and tuning each core individually will get you a better overall result and the AMD driver tells Windows which is the best core, you can even see it stared in Ryzen master, Please tell me you have AMD's chipset driver installed?

But its a lot of faffing about tuning and testing 8 individual cores, so i just use the all core optimiser. with it your optimal to the worst core but like you i'm too lazy.

Edit 12 in your case.

 

[TNA]

I know the feeling, some cores are better than others and tuning each core individually will get you a better overall result and the AMD driver tells Windows which is the best core, you can even see it stared in Ryzen master, Please tell me you have AMD's chipset driver installed?

But its a lot of faffing about tuning and testing 8 individual cores, so i just use the all core optimiser. with it your optimal to the worst core but like you i'm too lazy.

Edit 12 in your case.

Of course I am on AMD Chipset Drivers. I used to even update them each time new ones came up before. But these days I just update it from format to format.

 

[4K8KW10]

That CPU-Z benchmark very much depends on the temperature of the CPU, if you take your PC outside to 10°C, the results will improve.

My Ryzen 9 5900X after a cold start begins the multi-threaded test with 10,000... and then when the CPU warms up a bit, the result weakens to about 9,800-9,900.

 

[TNA]

That CPU-Z benchmark very much depends on the temperature of the CPU, if you take your PC outside to 10°C, the results will improve.

My Ryzen 9 5900X after a cold start begins the multi-threaded test with 10,000... and then when the CPU warms up a bit, the result weakens to about 9,800-9,900.

Yeah, that would explain the small difference.

On another note, I watched some Hardware Unboxed videos recently and you came to mind

You used to continually bang on about 6 cores no longer being enough and I would tell you that is not how it works, but you would keep banging on all the same. Well here you go, hopefully you learn something

 

[TNA]

I did one last test today before disabling and going back to stock. With this one it was -10 on all cores and +50MHz and got this:

In terms of percentage, still a very small difference. That is why I just disabled it for now as I would have to do a lot of testing to see if it is rock solid stable and be happy with it, which I can't be asked right now. With my 3600 it made sense to do so as I ended up with 200MHz more on all cores, but even more importantly it was using much less power at 1.275v vs stock which was much higher and struggled to boost to 4.2GHz, never mind stay at 4.4GHz at all times.

 

[4K8KW10]

Feel free to downgrade to a 6-core then.

 

[TNA]

Feel free to downgrade to a 6-core then.

So you watched and did not understand anything it seems. Not surprised.

How does those help your argument by the way? Helps mine. Lol.

 

[4K8KW10]

No, I did not watch, and understood everything. They said that more cores is always better.

Why are you insisting to keep you in my ignore list forever and ever?

 

[Joxeon]

Yeah, that would explain the small difference.

On another note, I watched some Hardware Unboxed videos recently and you came to mind

You used to continually bang on about 6 cores no longer being enough and I would tell you that is not how it works, but you would keep banging on all the same. Well here you go, hopefully you learn something

I seen that vid and tend to agree that spending less and upgrading more often is the way to go for futureproofing, I mean how futureproof is someone who spent almost £500 on an ryzen 7 1800X vs someone else who brought a 1600X then upgraded to a 5600X at a similar net spend.

 

[humbug]

I seen that vid and tend to agree that spending less and upgrading more often is the way to go for futureproofing, I mean how futureproof is someone who spent almost £500 on an ryzen 7 1800X vs someone else who brought a 1600X then upgraded to a 5600X at a similar net spend.

To some extent i agree with him but takes it too far, and he manipulates his audience, instead of properly informing people he does sometimes set up his reviews to prove his exaggerated truth and i'm not ok with that.

A lot of people bought Ryzen 3600 on his saying that a 9900K is only a few % faster in games, as if that difference was down to the CPU and not his testing methods, nothing of course could be further from the truth, the 9900K is a much faster CPU and those upgrading from their 2080TI to a 3080 are having to buy a 5800X when they didn't think they would need to.

And then he did it again with the Ryzen 3600 vs the 5600X pretending there is little difference between them while leaving off his charts the GPU that he used which was a 6700XT, an AMD version of a 2080TI, again...

 

[Joxeon]

To some extent i agree with him but takes it too far, and he manipulates his audience, instead of properly informing people he does sometimes set up his reviews to prove his exaggerated truth and i'm not ok with that.

A lot of people bought Ryzen 3600 on his saying that a 9900K is only a few % faster in games, as if that difference was down to the CPU and not his testing methods, nothing of course could be further from the truth, the 9900K is a much faster CPU and those upgrading from their 2080TI to a 3080 are having to buy a 5800X when they didn't think they would need to.

And then he did it again with the Ryzen 3600 vs the 5600X pretending there is little difference between them while leaving off his charts the GPU that he used which was a 6700XT, an AMD version of a 2080TI, again...

You see I actually find it more misleading when reviewers just show only 1080p benchmarks with the fastest GPU as it makes people who are only buying midrange cards think they need a stronger CPU when often it's an area they can save in.

An example of this is in a recent thread on here a guy buying a 5900X and X570 just to go with his 1660 super, I recommended a cheaper B550 and a 3600 but he thinks the 5900X is much faster and while it is, for his use case it won't be and the extra cash would have been better spent on a B550 + faster GPU.

 

[humbug]

You see I actually find it more misleading when reviewers just show only 1080p benchmarks with the fastest GPU as it makes people who are only buying midrange cards think they need a stronger CPU when often it's an area they can save in.

An example of this is in a recent thread on here a guy buying a 5900X and X570 just to go with his 1660 super, I recommended a 3600 but he thinks the 5900X is much faster and while it is for his use case it won't be and the extra cash would have been better spent on a B550 + faster GPU.

That happens too yes i know.

The proper way to inform someone about the CPU's is to show the true performance difference between all these CPU's so people can judge how many GPU generation upgrades its good for but also make it equally clear to people how well or not the CPU performs at IQ settings and resolutions they are likely to use with current GPU's

He doesn't do that, instead he presents his content as if you're looking at the actually CPU performance when in fact its the GPU, not the CPU and that's manipulative.

And what about E-Sports?, people with 240Hz and 360Hz panels, do they not exist at all?

But he's not alone, they all do it to some extent, apart from maybe Steve Burk, Anand Tech defiantly still know how to properly inform people.