AMD's "X" Nomenclature on Upcoming Ryzen Chips Related To XFR Feature

[newtekie1]

The only limit on CPUs currently is temperature. This isn't a locked TDP limit in any way. And this is really only to stop the chip from destroying itself. They don't do it to purposely limit a chip they sell as "overclockable".

 

[TheoneandonlyMrK]

The only limit on CPUs currently is temperature. This isn't a locked TDP limit in any way. And this is really only to stop the chip from destroying itself. They don't do it to purposely limit a chip they sell as "overclockable".

That's not the case with AMD fx processors ,you have to disable certain advanced power control settings to get beyond certain clocks and voltage points, you have to have the right board, as far as I can see you can properly overclock 3 Intel chips , the rest are directly limited in a different way, that's all ,you are not going much passed tdp thresholds bclk clocking after all , it's all relative.
I'm likely wrong in that indeed only X chips are rumoured to have the afxr auto overclocking anyway making a defined difference.

 

[newtekie1]

That's not the case with AMD fx processors ,you have to disable certain advanced power control settings to get beyond certain clocks and voltage points, you have to have the right board, as far as I can see you can properly overclock 3 Intel chips , the rest are directly limited in a different way, that's all ,you are not going much passed tdp thresholds bclk clocking after all , it's all relative.
I'm likely wrong in that indeed only X chips are rumoured to have the afxr auto overclocking anyway making a defined difference.

That isn't true of FX chips at all. The chips themselves have no hard locked limits other than temperature. Most of the motherboards have protection on the VRMs to stop them from frying if the power draw stays too high for too long, or the VRMs start to overheat. But, again, that isn't a hard lock built into the chip by the chip maker, that's the motherboard manufacturer protecting their motherboard from idiot users blowing them up.

Also, last I checked, Intel sells 10 overclockable processors, not 3.

 

[TheLaughingMan]

Are you trying to claim that 2 different chips that draw 95w of electricity would generate different amounts of heat ...

Yes. That is exactly what I am saying. Two different chips with two different manufacturing methods will dissipate different amounts of heat based on its design, heatsink, efficiency, etc. etc. Its just like comparing two PSUs with different efficiency ratings with the same Wattage output. One will draw more power and generate more heat. All you have done in your scenario you have locked the input voltage which still means one chip will use more power and generate more heat based on these factors.

 

[newtekie1]

Yes. That is exactly what I am saying. Two different chips with two different manufacturing methods will dissipate different amounts of heat based on its design, heatsink, efficiency, etc. etc. Its just like comparing two PSUs with different efficiency ratings with the same Wattage output. One will draw more power and generate more heat. All you have done in your scenario you have locked the input voltage which still means one chip will use more power and generate more heat based on these factors.

That isn't how physics work. The power consumed(energy) must go somewhere. It either has to go to movement, light, or heat. In the case of processors, there are no moving parts, there is no light generated, so all of the power consumed is put out as heat. If two processor both consume 95w of power, then they are dissipating the same amount of heat. They both might be able to do different amount of calculations with the power they are consuming, but if they are both consuming 95w they are both putting out the same amount of heat. The design, heatsink, and efficiency don't change physics.

 

[TheLaughingMan]

That isn't how physics work. The power consumed(energy) must go somewhere. It either has to go to movement, light, or heat. In the case of processors, there are no moving parts, there is no light generated, so all of the power consumed is put out as heat. If two processor both consume 95w of power, then they are dissipating the same amount of heat. They both might be able to do different amount of calculations with the power they are consuming, but if they are both consuming 95w they are both putting out the same amount of heat. The design, heatsink, and efficiency don't change physics.

Where does it get the power to change bit in memory. Some form of electrical energy is needed to send a signal to cache, memory, HDD, video card, etc. to change a bit. If all energy into the CPU is lost to heat, where is it getting power to communicate with the rest of the computer?

 

[medi01]

Where does it get the power to change bit in memory. Some form of electrical energy is needed to send a signal to cache, memory, HDD, video card, etc. to change a bit. If all energy into the CPU is lost to heat, where is it getting power to communicate with the rest of the computer?

Electric energy flows inside the PC case from power socket over power cables.
Its only way out is heat or (ignorable, since it is way to small) EM radiation.
Nothing else.

Whenever that "some form of electrical energy" (not that we know many forms of it, but fuck it) is used to send a signal to cache/memory/whatever, it ends up either emitting a tiny bit of EM waves, or just heats things up ("makes molecules move faster", not 100% correct, but very close)


Bottom line: when 95W TDP rated chip consumes roughly as much as 145W TDP chip and total consumption of systems is actually about 180 watt, the 95W TDP rating is clearly BS.

What I was asking was why on earth did AMD need such bogus TDP numbers, do they really think they can get away with when nearly all reviewers do power consumption tests?

 

[TheoneandonlyMrK]

Electric energy flows inside the PC case from power socket over power cables.
Its only way out is heat or (ignorable, since it is way to small) EM radiation.
Nothing else.

Whenever that "some form of electrical energy" (not that we know many forms of it, but fuck it) is used to send a signal to cache/memory/whatever, it ends up either emitting a tiny bit of EM waves, or just heats things up ("makes molecules move faster", not 100% correct, but very close)


Bottom line: when 95W TDP rated chip consumes roughly as much as 145W TDP chip and total consumption of systems is actually about 180 watt, the 95W TDP rating is clearly BS.

What I was asking was why on earth did AMD need such bogus TDP numbers, do they really think they can get away with when nearly all reviewers do power consumption tests?

Tdp is as the laughing man says and due to process variations you get different leakage with different chips because of physics.
Two chips of the same design can pull different power and output different heat amounts for a given clock.

All chip companies quote tdp or how would you know how much cooling to fit, except Intel who switched to Sdp system designed power.

That's a fact.
@newtekie1 I've given up mate I've got a fx it doesn't clock high with awpm switched on...

 

[medi01]

Two chips of the same design can pull different power and output different heat amounts for a given clock.
That's a fact.

That's not what he says.
He says you can pull more, yet produce less heat.
https://www.techpowerup.com/forums/...ted-to-xfr-feature.230551/page-3#post-3600730

 

[TheoneandonlyMrK]

That's not what he says.
He says you can pull more, yet produce less heat.
https://www.techpowerup.com/forums/...ted-to-xfr-feature.230551/page-3#post-3600730

If one chips made with less defects or better in general (cleaner interconnects etc)that's a possibility as it's leakage will be less making it more efficient..

 

[newtekie1]

If one chips made with less defects or better in general (cleaner interconnects etc)that's a possibility as it's leakage will be less making it more efficient..

Again, that isn't what he is saying. He is saying it can consume more power, but produce less heat. That is impossible.

@newtekie1 I've given up mate I've got a fx it doesn't clock high with awpm switched on...

I've got several FX chips. I run an 8350 24/7. It will run whatever clock and voltage I tell it to. I've even killed a 8320e by giving it too much voltage on accident. The chips don't care, the motherboard might, but the chips don't.

 

[TheoneandonlyMrK]

Again, that isn't what he is saying. He is saying it can consume more power, but produce less heat. That is impossible.



I've got several FX chips. I run an 8350 24/7. It will run whatever clock and voltage I tell it to. I've even killed a 8320e by giving it too much voltage on accident. The chips don't care, the motherboard might, but the chips don't.

So post me a screenshot of it hitting 5.5 then with all CPU features bar hpc on and I'll maybe believe you

 

[newtekie1]

So post me a screenshot of it hitting 5.5 then with all CPU features bar hpc on and I'll maybe believe you

I don't have the cooling for that. My 8350 is at 4.6GHz right now, and overpowers the 120mm rad I have cooling it as it is. Of course, if it was TDP locked, adding better cooling wouldn't matter, because the amount of heat it outputs would never get above the limit...but we all know that doesn't happen. Well, everyone but you apparently.

You are more than welcome to post some proof that they have a TDP hard lock though.

 

[TheoneandonlyMrK]

I don't have the cooling for that. My 8350 is at 4.6GHz right now, and overpowers the 120mm rad I have cooling it as it is. Of course, if it was TDP locked, adding better cooling wouldn't matter, because the amount of heat it outputs would never get above the limit...but we all know that doesn't happen. Well, everyone but you apparently.

You are more than welcome to post some proof that they have a TDP hard lock though.

So what do you know about hitting its limits then , you hit your coolings limit, I didn't , with two 360 rads and a 120 I hit the chips limit not my coolings like you.
It's because I wasn't thermally restricted that I Know this they don't advertise it as a feature, shit no one knows what the top temp these can run at I've heard 62, 72 all sorts mine stays below 60.
Yet as I'm saying with awpm on in BIOS you won't get to 5.5 with it off I have without crashing with a kernel power error.
Check the many terminal run oc fx CPU threads on here you will find it commonly noted for max clocks turn off all the CPU features like cooln n quite ,hpc and apwm, usually ok to leave on upto 4.8 and your at 4.6.



every cpu chip has a thermal limit and if enabled it will shut your pc off ifff triggered.
And it's above the tdp not at it ,normally between 95 105 degrees.

 

[newtekie1]

So what do you know about hitting its limits then , you hit your coolings limit, I didn't , with two 360 rads and a 120 I hit the chips limit not my coolings like you.

That is my situation now, not in the past. I've scaled down greatly. I had the 8320e under DICE, it took all the voltage and clocks I gave it, until it died.

If what you say is true. Show me your chip throttling due to a TDP limit.

 

[TheoneandonlyMrK]

That is my situation now, not in the past. I've scaled down greatly. I had the 8320e under DICE, it took all the voltage and clocks I gave it, until it died.

If what you say is true. Show me your chip throttling due to a TDP limit.

I never said it throttled.
With those settings off you can do what you want to the chip.
And Tdp isn't a limit more a fact.

Thermal design power in chip terms is used to rate the max thermal dissipation required during use.

After that it explains nothing about how a chip will work and I never said it did.

AMD still use tdp that same way

Intel do sort of even with Sdp.

Because of modern chips varying core frequency (note I'm not mentioning turbo) from low to high they all control there own performance in the moment and you have to bypass something in all CPUs in order to maintain a top clock ,yes or no?

So what happens when some tool tries to get 5ghz on the stock ally hsf
The CPU sees it's about to die and shuts down ,you can disable that and make it just die again yes or no?

Since the whole motherboard is being drawn into the CPU it is inevitable that the CPU will constrain control over itself and get more sensors and critical ,or not limits.

Getting back to RyZen for me it's fine that X editions have a higher tdp then none X

It's also fine that the non X will overclock but has a limit I don't see an issue Intel does the same but officially bans overclocking non endorsed chips and even frowns at OEMs trying to circumvent this.

 

[newtekie1]

Yes, I know what the TDP rating is, I know what TDP is. Do you know what "hard locked" TDP means? Because obviously you don't even though you've been claiming this whole time AMD FX chips have it... If you can just turn off a few settings in the BIOS to bypass the limit, then it isn't a hard limit.

I'll say it one more time, and that's it, I don't have a problem with the lower TDP, my problem is the lower TDP "Hard Limit".

 

[TheoneandonlyMrK]

Yes, I know what the TDP rating is, I know what TDP is. Do you know what "hard locked" TDP means? Because obviously you don't even though you've been claiming this whole time AMD FX chips have it...

I'll say it one more time, and that's it, I don't have a problem with the lower TDP, my problem is the lower "hard locked" TDP.

ok so I miss understood or interpreted you I apologise
But I don't know what you mean then because I never mentioned fx chips having a hard lock I mearly said that fx processors have a early version of it that limits overclocking if enabled.
To be fair we are only speculating at this point and I might well have started something for nothing if they don't have any limits but they will I think.

Messing with AMD overdrive software for CPUs displays some of the opportunity missed and yet to be reaped since it's very good in some ways ,going beyond what can be done just in BIOS with cpuvid tweaking etc.
Drag all that into the chip and they could pretty much configure and run themselves.
And actively limit max to be within its bin rating which is effectively what would could happen.
Again speculating I should just shut up but I'm eager for reviews.

 

[newtekie1]

ok so I miss understood or interpreted you I apologise
But I don't know what you mean then because I never mentioned fx chips having a hard lock I mearly said that fx processors have a early version of it that limits overclocking if enabled.

I don't see how. My very first post quoted "65w Hard Limit" and I said I hope that isn't true. I was very clear what I was talking about. You argued with me about it, and eventually claimed FX chips already have it. When I said the only limit built into the chip itself was a temperature limit, you said:

That's not the case with AMD fx processors

The fact is, there are no other limits built into the FX chips themselves. All other limits are imposed by the motherboards, not the FX chips themselves. If you can disable it, it isn't a limit, it is just a safe guard. A limit is something you can not bypass. There is no option to turn it off. That is what concerns me about a hard limit in TDP. That implies it can not be bypassed, and AMD has purposely crippled overclocking on chips they market as overclockable with unlocked multipliers.

I even explained this to you already back in post 38. And your response was that the e series FX chips TDP was locked. That isn't true.

 

[TheoneandonlyMrK]

I don't see how. My very first post quoted "65w Hard Limit" and I said I hope that isn't true. I was very clear what I was talking about. You argued with me about it, and eventually claimed FX chips already have it. When I said the only limit built into the chip itself was a temperature limit, you said:


The fact is, there are no other limits built into the FX chips themselves. All other limits are imposed by the motherboards, not the FX chips themselves. If you can disable it, it isn't a limit, it is just a safe guard. A limit is something you can not bypass. There is no option to turn it off. That is what concerns me about a hard limit in TDP. That implies it can not be bypassed, and AMD has purposely crippled overclocking on chips they market as overclockable with unlocked multipliers.

I even explained this to you already back in post 38. And your response was that the e series FX chips TDP was locked. That isn't true.

I'll throw some quotes your way when I get home I can't be assed on a phone.
But in part you turn off CPU features in the CPU section and motherboard parts in other areas of the BIOS.

 

[newtekie1]

I'll throw some quotes your way when I get home I can't be assed on a phone.
But in part you turn off CPU features in the CPU section and motherboard parts in other areas of the BIOS.

Again, if you can turn them off, they aren't hard limits. So it doesn't really matter.

 

[RejZoR]

Hard locked TDP and overclocking able CPU in one sentence doesn't make any sense. The TDP they measured was most likely up to a boost clock. Anything beyond that in form of XFR is not counted as that's considered overclocking already. By that alone it'll exceed the stated TDP for given CPU.