CCX Overclocking Tool for AMD CPUs Updated With New Features

[Tatty_Two]

no,it's called "calling it as it is".
and this is mostly on people who hyped zen 2 on froums,not amd themselves.amd never promised anything remotely close to what people kept insisting zen 2 would be.
using a dangerous tool to prove a point that ryzen can do 4.5ghz is silly,just enjoy zen 3000 for what is,an excellent all round cpu with great energy efficiency.don't ruin it with a stupid tool.

I agree but in this thread the Hype train/expectations/reality and flaws perceived or not is off topic which is probably their point.

 

[Manu_PT]

It's called Troll.I never do that , even in Intel/Nvidia.He tried hard in every AMD topic and this is what you see. Doesn't Matter Story is true or Wrong also you too.

You know why? Because Im sure you are not dealing with buggy Bios and CPUs idling at 60° and voltages going up to 1,5v by opening youtube, random restarts, insane high temps, ALL at STOCK, everyday

If you had customers questioning the company where you work or questioning staff competence, maybe you would have same opinion.

This launch is a mess, is nothing close to what was rumoured by fanboys. I have the right to my opinion if I see threads like this, about a tool that is risky, just to get 100mhz more out of a CPU line up that barely overclocks.

Implying that the stock processor is unstable in anyway way and manual overclocks requires no testing.
This is false for both Intel and AMD.
As usually all overclocking are done on your own risk. This does not change no matter what tool you are using.

Except that zen2 is unstable even at stock. That's the problem.

 

[Digital Dreams]

I have a job. I rather spend 50€/100€ more on something stable, thanks.

So...only unemployed people buy AMD?
Also I wasn't aware AMD was unstable. You're trying way too hard.

 

[cucker tarlson]

So...only unemployed people buy AMD?
Also I wasn't aware AMD was unstable. You're trying way too hard.

I think he meant the tool not the cpu/company.

 

[kapone32]

You know why? Because Im sure you are not dealing with buggy Bios and CPUs idling at 60° and voltages going up to 1,5v by opening youtube, random restarts, insane high temps, ALL at STOCK, everyday

If you had customers questioning the company where you work or questioning staff competence, maybe you would have same opinion.

This launch is a mess, is nothing close to what was rumoured by fanboys. I have the right to my opinion if I see threads like this, about a tool that is risky, just to get 100mhz more out of a CPU line up that barely overclocks.



Except that zen2 is unstable even at stock. That's the problem.

That is the first I am hearing about Ryzen3 CPUs being unstable. This is one of the reasons why I like to test components myself when affordable.

 

[cucker tarlson]

That is the first I am hearing about Ryzen3 CPUs being unstable. This is one of the reasons why I like to test components myself when affordable.

I don't think it's unstable.
oh wait,there was some agesa fail,but that was quickly pulled.

 

[Hardware Hound]

Reading through the comments, I'm still a little unsure how the CCX overclocking tool could do that much harm. I have a 3600X, and it looks like there are 3 cores per CCX. If one of those modules can only hit 4.4GHz, that would limit the whole CPU. The other module might hit 4.5GHz, so I would have 3 cores at 4.4, 3 cores at 4.5. Sounds like a pretty sweet deal. (Hypothetically speaking.)

I would be supplying the same voltage across the cores it seems, so wouldn't it be up to cooling to handle the overclock? I'm just not understanding how this overclocking tool could do something to burn out a CPU any more than normal overclocking with raising the multiplier. Sure, if a crazy voltage spike in excess of 2v hit the cores, that might do it. My experience has been that heat is the main concern. It sounds like the CCX tool isn't doing anything special to voltage, just grouping cores by CCX. Some modules will bin higher within the same CPU. Setting the voltage manually and being careful with LLC should keep the voltage safe right?

I'm genuinely asking though since there's a lot on an engineering level I don't understand.

 

[Fr3ak]

Also, you can overclock per CCX or individual cores using Ryzen Master, which I have to use for overclocking anyway because Gigabyte forgot (or never knew) how to program a decent BIOS that actually sets values you selected in the BIOS instead of just causing endless boot loops. And it would be nice if I could use my M.2 again, it's been a couple of weeks already. How hard can it be? Other than people claiming to hit higher clocks without Ryzen Master installed (which I cannot verify), I also don't quite get the point of using yet another software tool.

 

[IceShroom]

10 months ago, 5ghz 8 cores are coming, entry level, half of Intel price. Rip Intel

2 months ago, no 5ghz, thats unrealistic, are you dumb? Amd just needs 4,5ghz to defeat Intel on every scenario anyway. Rip Intel

Present day, let's spend hours on bios, ryzen master, special toolsz to squeze 100mhz, as our chips cant even sustain 4,2ghz all core turbo.

This is the usual cycle on Amd. I think all of this is pathetic. I have a job. I rather spend 50€/100€ more on something stable, thanks.

You know that was a speculation by someone not official word from AMD. Maybe do some reading on news other than reading Nvidia/Intel marketing memo supplied to you.

 

[Lionheart]

Well judging from these comments, A well known overclocker made this tool (Pretty awesome btw) but a well known AMD engineer is warning you to use it at your own risk, Annnnd Manu_PT & Cucker need to get a room. Gotcha thanks.

 

[eidairaman1]

Delete this application and forget about existence, you will kill your processors!
This is hacking FIT limits, the last line of protection for the processor

You may want to talk to shamino and amd about that then...

 

[Vya Domus]

You know why? Because Im sure you are not dealing with buggy Bios and CPUs idling at 60° and voltages going up to 1,5v by opening youtube, random restarts, insane high temps, ALL at STOCK, everyday

If you had customers questioning the company where you work or questioning staff competence, maybe you would have same opinion.

This launch is a mess, is nothing close to what was rumoured by fanboys. I have the right to my opinion if I see threads like this, about a tool that is risky, just to get 100mhz more out of a CPU line up that barely overclocks.



Except that zen2 is unstable even at stock. That's the problem.

I would question your competence too, especially if I knew you waste your time spamming every AMD related thread with this FUD garbage instead of performing whatever job your supposed to do.

 

[TheGuruStud]

AMD should just improve their OC features (not talking about overclockability headroom) to be onpar with Intel. When I had the 1700x, I was forced to use that ghetto p-state overclocking to be able to downclock the CPU at idle.

That's a MB problem unrelated to cpu.

 

[Hardware Hound]

I find it funny after this long since the FX years that people are now talking about Intel being at 5GHz as if that number is important. Remember the 9590 and the whole, "First CPU to reach 5GHz" garbage back then. If you never update an architecture or process node, it's easier to push higher frequencies. Doesn't equate to full performance improvements though.

Intel has been on the same process node for quite some time. In spite of this, AMD is nearly matching single core performance at a lower frequency. Ryzen will probably need a couple more years to mature enough to push the 5GHz boundary. Maybe by then, Intel will have the new architecture and process node to give a solid leap at performance, which I look forward to. In the end, it's price and performance. Personally, I'm super happy to have an option like the 3600. Great price and a super versatile chip for gaming and workloads.

 

[Tomorrow]

I find it funny after this long since the FX years that people are now talking about Intel being at 5GHz as if that number is important. Remember the 9590 and the whole, "First CPU to reach 5GHz" garbage back then. If you never update an architecture or process node, it's easier to push higher frequencies. Doesn't equate to full performance improvements though.

Intel has been on the same process node for quite some time. In spite of this, AMD is nearly matching single core performance at a lower frequency. Ryzen will probably need a couple more years to mature enough to push the 5GHz boundary. Maybe by then, Intel will have the new architecture and process node to give a solid leap at performance, which I look forward to. In the end, it's price and performance. Personally, I'm super happy to have an option like the 3600. Great price and a super versatile chip for gaming and workloads.

^ This

5Ghz has no meaning if it does not actually improve performance. Yes AMD might bet at 4.2 but that is almost as fast as Intel's 5Ghz.
Actually Intel tried this with their Pentium 4 era. Going for massive clock speeds instead of per clock cycle performance. By all account's it was a massive failiure just like Bulldozer.

3900X is on average only 5% slower than 9900K in games. Even when BOTH are overclocked. In edge cases 9900K migh be 20% faster but those are rare.
Why would i take a CPU that is 5% faster in one task and loses 30-50% in other tasks, has no platform upgrade path, has more severe security issues but costs the same?

 

[eidairaman1]

I find it funny after this long since the FX years that people are now talking about Intel being at 5GHz as if that number is important. Remember the 9590 and the whole, "First CPU to reach 5GHz" garbage back then. If you never update an architecture or process node, it's easier to push higher frequencies. Doesn't equate to full performance improvements though.

Intel has been on the same process node for quite some time. In spite of this, AMD is nearly matching single core performance at a lower frequency. Ryzen will probably need a couple more years to mature enough to push the 5GHz boundary. Maybe by then, Intel will have the new architecture and process node to give a solid leap at performance, which I look forward to. In the end, it's price and performance. Personally, I'm super happy to have an option like the 3600. Great price and a super versatile chip for gaming and workloads.

9590 single core turbo, my 8350 5.0 all cores on air. Fx had half resources compared to phenom 2.

Intel has done nothing but sandbag the last decade now they're dragging their feet plus they left their arch unsecure thinking it would go unnoticed.

^ This

5Ghz has no meaning if it does not actually improve performance. Yes AMD might bet at 4.2 but that is almost as fast as Intel's 5Ghz.
Actually Intel tried this with their Pentium 4 era. Going for massive clock speeds instead of per clock cycle performance. By all account's it was a massive failiure just like Bulldozer.

3900X is on average only 5% slower than 9900K in games. Even when BOTH are overclocked. In edge cases 9900K migh be 20% faster but those are rare.
Why would i take a CPU that is 5% faster in one task and loses 30-50% in other tasks, has no platform upgrade path, has more severe security issues but costs the same?

Plus costs a ridiculous amount of money

 

[lexluthermiester]

Shamino is a very well known overclocker since 2000. This tool is a very good addition for advanced users who want to overclock. Also make point that if people want to be an overclocker then they must have ways to log motherboard and cpu voltages to see any danger that could cause to their cpus.

Also remember that overclocking could include all types of hacks, your own risk if you decide to use it.

That's kinda what my thinking was. Of course we need to verify that it's actually "Shamino".

 

[Jism]

Um, you're joking right?

All Ryzen CPUs have a FIT or simply Silicon health checking feature. This allows PBO or XFR to always seek the highest possible boost without making things unstable, overvolting or putting too much current through the CPU. Since the CPU is made on a very small proces node, it doesnt take much too much current to kill it. i.e when overvolting and running 24/7 on 1.40 to 1.45v shows signs of degradation in barely 6 to 3 months.

So yeah you could overclock a CCX or independent core, but at the cost of something else. If your going to bypass the FIT it takes one bad run on any heavy workload and your killing your CPU. These chips are not FX´s that eat 1.55 to 1.65v on water all day. These are chips with a much thinner internal wiring that a bit of current would simply pop that wiring.

This is why you dont see a all core 4.3GHz boost on a 2700x for example, when running IBT, but a more conservative 4.1GHz all core boost (when cooling is proper). This is just FIT limits while respecting the silicon limits.

Nobody on the net done a proper research except for the stilth related to FIT and what it is exactly. In a technical discussion on anandtech he wrote a few lines on the FIT on AMD cpu´s. If you wanna risk blowing your CPU be sure to use that above tool. Because FIT is being thinkered into something else just like a resistor on a temperature sensor that is mislead different. You can get very strange behaviour and eventually blow your CPU.

AMD has done a great job for the masses to bring a intelligent boost state to its CPUs but a terrible job for any of the manual overclockers out there.

 

[Hardware Hound]

All Ryzen CPUs have a FIT or simply Silicon health checking feature. This allows PBO or XFR to always seek the highest possible boost without making things unstable, overvolting or putting too much current through the CPU. Since the CPU is made on a very small proces node, it doesnt take much too much current to kill it. i.e when overvolting and running 24/7 on 1.40 to 1.45v shows signs of degradation in barely 6 to 3 months.

So yeah you could overclock a CCX or independent core, but at the cost of something else. If your going to bypass the FIT it takes one bad run on any heavy workload and your killing your CPU. These chips are not FX´s that eat 1.55 to 1.65v on water all day. These are chips with a much thinner internal wiring that a bit of current would simply pop that wiring.

This is why you dont see a all core 4.3GHz boost on a 2700x for example, when running IBT, but a more conservative 4.1GHz all core boost (when cooling is proper). This is just FIT limits while respecting the silicon limits.

Nobody on the net done a proper research except for the stilth related to FIT and what it is exactly. In a technical discussion on anandtech he wrote a few lines on the FIT on AMD cpu´s. If you wanna risk blowing your CPU be sure to use that above tool. Because FIT is being thinkered into something else just like a resistor on a temperature sensor that is mislead different. You can get very strange behaviour and eventually blow your CPU.

AMD has done a great job for the masses to bring a intelligent boost state to its CPUs but a terrible job for any of the manual overclockers out there.

This guy, right here! Thanks for clarifying on what the FIT is and clearing up that CCX overclocking does indeed bypass it.

Basically, a CPU is a lot like a fuse in that regard. A fuse uses a wire that is just small enough, that if too much voltage or amperage hits it, it blows the wire. Of course with a fuse, it's protecting something else. With a CPU, you just lost some money. The wire analogy may not be 100% accurate, but it does help clarify what is happening.

I hope that better overclocking support comes with further generations. Currently, I understand catering to the masses more than the overclockers since the company has to make profits. Hope newer releases change that though.

 

[Jism]

The AMD FX had a limit too, with special high end boards you where able to override the AMD specification on the FXs current limit. By doing that you allowed to CPU to put heavy workloads at higher clocks then most of the people could do. This way you could overclock your CPU all the way up to 5Ghz and have a staggering 250W CPU power consumption on tests like Intel Burn Test. Ive saw alot of people facing the same issues, not getting their CPU stable beyond 4.4GHz. The trick was to turn off a core to determine if it was a current limit or not set by AMD or the motherboard.

I think the same applies to ryzen, where AMD this time hardcoded the current limit in the FIT. This current limit can be bypassed by proberly this tool, but at the cost of putting your CPU in danger. It really takes once pounding workload related to current and gone is your CPU. Its irreversible damage. Yes a CPU is eventually all wiring, and that wiring has a certain limit before it blows. AMD designed the FIT so it would always stay within safe limits regarded to XFR / PBO. Perhaps people could archieve a bit higher clocks when they would bypass the current limit, but nobody knows exactly when or where its going to blow.

Again: i seen a guy degrading his 2700x in barely a few months by running on 1.45v. This proces normally would take a few years. This is why its so dangerous going beyond the recommended 24/7 of 1.34v on 2x00 series and 1.37v on 3x00 series. The minimum requirement for any good cooling would be 360 radiators and 6 fans or so. The usual 240 rads are capable for a OC, but not the extremer OC´s without hitting temperature limits.

Ive learned that the 2700x performs best when kept under 60 degrees. It will hold the boost state way longer and even all core on 4.1Ghz for a longer term. Its all about FIT, cooling and motherboard VRM capabilities.

 

[Hardware Hound]

Yeah, I see this being useful for overclocking competitions, but those guys can top 2v when using LN2. They also don't care about degradation at that point. I will not be using this on my test bench for sure. Don't have the money to risk damaging a CPU.

 

[Nkd]

I'm really disappointed with the OC potential of ryzen3000. basically it just approached Intel 14nm and I would like to know or will be better with the R9 3950 model where they boasted with 5.2 gh OC working . This is not a gpu, although the intel coffee like it probably has an intergrated virtue and achieves on smaller resolutions a lot of advantage but at 4k the differences are almost alignment . Even ram contability lists have become shorter and reach 3433 Hz 2 channels and 2966Hz in all four. (PRESTIGE x570 board from MSI list). so I would briefly say that these little corrections do not give you confidence .The truth is different from the advertised and the regular price for ryzen R9 3900X is 777 € and online 621.66 € should bee 500$ which is much cheaper . I am very disappointed with the OC capability and I think that this is a very poor quality silicone where the main culprit for the results .This is my opinion .

Yea 15-20% difference in IPC is very disappointing from previous gen. Blame TSMC for clocks not AMD. They kept lowering their targets as the process went on, the extra 10% got pushed to 7nm+ so AMD couldn't really do much about clocks. But they nailed it on IPC. Its like running zen+ close to 5ghz. Zen 3 will likely be north of 4.5ghz all core OC, thats where we are at.

Not to mention I am running 3900x at 4.4ghz all core stress tested and stable.

The AMD FX had a limit too, with special high end boards you where able to override the AMD specification on the FXs current limit. By doing that you allowed to CPU to put heavy workloads at higher clocks then most of the people could do. This way you could overclock your CPU all the way up to 5Ghz and have a staggering 250W CPU power consumption on tests like Intel Burn Test. Ive saw alot of people facing the same issues, not getting their CPU stable beyond 4.4GHz. The trick was to turn off a core to determine if it was a current limit or not set by AMD or the motherboard.

I think the same applies to ryzen, where AMD this time hardcoded the current limit in the FIT. This current limit can be bypassed by proberly this tool, but at the cost of putting your CPU in danger. It really takes once pounding workload related to current and gone is your CPU. Its irreversible damage. Yes a CPU is eventually all wiring, and that wiring has a certain limit before it blows. AMD designed the FIT so it would always stay within safe limits regarded to XFR / PBO. Perhaps people could archieve a bit higher clocks when they would bypass the current limit, but nobody knows exactly when or where its going to blow.

Again: i seen a guy degrading his 2700x in barely a few months by running on 1.45v. This proces normally would take a few years. This is why its so dangerous going beyond the recommended 24/7 of 1.34v on 2x00 series and 1.37v on 3x00 series. The minimum requirement for any good cooling would be 360 radiators and 6 fans or so. The usual 240 rads are capable for a OC, but not the extremer OC´s without hitting temperature limits.

Ive learned that the 2700x performs best when kept under 60 degrees. It will hold the boost state way longer and even all core on 4.1Ghz for a longer term. Its all about FIT, cooling and motherboard VRM capabilities.

This guy gets it!

 

[timta2]

I agree but in this thread the Hype train/expectations/reality and flaws perceived or not is off topic which is probably their point.

To play Devil's Advocate, how can a comment about Ryzen overclocking expectations/potential be "off topic", in a thread about Ryzen overclocking?

 

[Mussels]

To play Devil's Advocate, how can a comment about Ryzen overclocking expectations/potential be "off topic", in a thread about Ryzen overclocking?

the moment someone goes "hurr durr AMD is shit buy intel like i did cause it makes my e-dick hard knowing i'm better than you" is a good sign its off-topic and worthy of deletion.

 

[lexluthermiester]

the moment someone goes "hurr durr AMD is shit buy intel like i did cause it makes my e-dick hard knowing i'm better than you" is a good sign its off-topic and worthy of deletion.

Thank You. I needed that laugh!