4,2XFR on threadripper. Higher binned/new stepping chips) ?

[The Benjamins]

Just now, Misanthrope said:

Can they rehuse the same die on a different socket chip though? I thought that wasn't really a thing but well that'd make sense.

all 3 stacks use the same die. they just put them in a different package and disable sections for its use case.

[Adaa]

3 minutes ago, Misanthrope said:

Can they rehuse the same die on a different socket chip though? I thought that wasn't really a thing but well that'd make sense.

The R7 and R5 and R3 are same socket, not sure what your referring too.

 

If you referr to threadripper, i dont think those chips are being devoted to normal Ryzens, unless they are litterly 2 glued togehter R7 chips as Intel claims.

[Misanthrope]

2 minutes ago, Adaa said:

The R7 and R5 and R3 are same socket, not sure what your referring too.

 

If you referr to threadripper, i dont think those chips are being devoted to normal Ryzens, unless they are litterly 2 glued togehter R7 chips as Intel claims.

 

3 minutes ago, The Benjamins said:

all 3 stacks use the same die. they just put them in a different package and disable sections for its use case.

To clarify: I was talking about how only 3 Threadripper SKUs have been announced and if this is just no improvement to manufacturing and just really nice bins, there'd be either a lot of the 8/16 Threadripper SKU floating around or many wasted chips: Binning works better on Ryzen cause there's 8 different SKUs all descending so there's 7 places outside of the 1800x for a chip to land vs just 2 places for a 1950x to land.

[Adaa]

1 minute ago, Misanthrope said:

 

To clarify: I was talking about how only 3 Threadripper SKUs have been announced and if this is just no improvement to manufacturing and just really nice bins, there'd be either a lot of the 8/16 Threadripper SKU floating around or many wasted chips: Binning works better on Ryzen cause there's 8 different SKUs all descending so there's 7 places outside of the 1800x for a chip to land vs just 2 places for a 1950x to land.

Yes i think, The Benjamins, are rigth. I think they can produce the modules seperately, so as long as a 4core CCX, works at a certain speed, they can missmatch them together left and rigth, to combine it all into the 4-8-12-16 parts we know. 

Theres likely some more work on putting them into the Threadrippers compared to normal Ryzens, but they also do take a price premium for those Threadrippers, so that do make sence. I bet AMD have a minimum they have to sell each 4ghz capable 4core CCX module for to make a buck, and the same for each 3,6 or 3,8 capable module. this way they can minimize spill, and predict the production loss and costs, of each xxxx mhz module.

[The Benjamins]

2 minutes ago, Misanthrope said:

 

To clarify: I was talking about how only 3 Threadripper SKUs have been announced and if this is just no improvement to manufacturing and just really nice bins, there'd be either a lot of the 8/16 Threadripper SKU floating around or many wasted chips: Binning works better on Ryzen cause there's 8 different SKUs all descending so there's 7 places outside of the 1800x for a chip to land vs just 2 places for a 1950x to land.

well they bin them for ryzen, TR and epyc. they all use the same die, it all comes from the same bucket. the same die used in the Epyc 7601 can be used in a R3 1200.

 

so again I don't see how it is wasteful, if they have to many high end chips for lets say the 1950x then demand needs, they can just put them in 1800x or 1700x or 1700, ect.

[tom_w141]

1 hour ago, Ryan_Vickers said:

I still don't understand the point of XFR though, so, it has a base clock, and a boost clock

Its like turbo boost 2 and 3. Just think of it that way. XFR is a couple of cores at a higher frequency aka turbo boost 3.0

[tom_w141]

1 hour ago, Adaa said:

why Ryzen dosnt do dynamical overclocking/boosting on more than 1 core,

XFR is 2 cores on my R7. So like turbo boost 3.0

[vanished]

1 minute ago, tom_w141 said:

Its like turbo boost 2 and 3. Just think of it that way. XFR is a couple of cores at a higher frequency aka turbo boost 3.0

turbo boost 3 is another thing I don't understand.  normal turbo boost already had a range of speeds: for example, a 4770k is reported as having a base clock of 3.5 GHz and a boost clock of 3.9 GHz, but that 3.9 is only for one core.  if two cores are used, it's 3.8, and if 3 or 4 cores are used, it's only 3.7.  ie, regular turbo boost already would go higher on fewer cores, and not as high when boosting many, so why did they need another technology to take a few cores extra far?  Wasn't that already a function of regular turbo boost?

[Adaa]

3 minutes ago, tom_w141 said:

XFR is 2 cores on my R7. So like turbo boost 3.0

1-2 cores , same/same. 

 

i wanted a boost 3.0 with 3-5 cores, so games could actually keep the higher clocks when ever all cores was not used. ;-) 

[tom_w141]

2 minutes ago, Ryan_Vickers said:

turbo boost 3 is another thing I don't understand.  normal turbo boost already had a range of speeds: for example, a 4770k is reported as having a base clock of 3.5 GHz and a boost clock of 3.9 GHz, but that 3.9 is only for one core.  if two cores are used, it's 3.8, and if 3 or 4 cores are used, it's only 3.7.  ie, regular turbo boost already would go higher on fewer cores, and not as high when boosting many, so why did they need another technology to take a few cores extra far?  Wasn't that already a function of regular turbo boost?

Its not clear I know just saying its present on both teams. Just XFR is a couple cores getting pushed further like turbo boost 3.0

[tom_w141]

Just now, Adaa said:

i wanted a boost 3.0 with 3-5 cores, so games could actually keep the higher clocks when ever all cores was not used. ;-) 

Well unfortunately neither team offers that  

[DXMember]

1 hour ago, Ryan_Vickers said:

 

 

Ah.  Well it's still an unnecessary distinction, just, the line that doesn't need to exist is the one between boost and XFR.  People shave gotten so used to always running at max turbo (on Intel) that I think they've forgotten that the idea of it was to boost as far as it can, provided enough thermal, current, voltage, etc. headroom.  Sounds like that's what XFR is too, why does it need to be differentiated from "boost clock"?

Well Intel also has idle clock, base clock, boost clock, tubo boost clock and tubo boost max clock

[tom_w141]

1 hour ago, DXMember said:

looking forward to all core 4.2 overclock

outside of LN2 no chance.

 

Maybe with phase change cooling and a voltage unsafe for daily use

[DXMember]

3 minutes ago, tom_w141 said:

outside of LN2 no chance

Lucky_n00b (Indonesia) Pushes AMD Threadripper 1950X to 5,187MHz, Breaks 4K in Cinebench R15

http://hwbot.org/newsflash/4691_lucky_n00b_(indonesia)_pushes_amd_threadripper_1950x_to_5187mhz_breaks_4k_in_cinebench_r15

 

btw, AMD's boost clock of 4GHz is for four cores, and XFR also goes for the four cores

[TVwazhere]

1 hour ago, Adaa said:

I dont think any media reviews has talked about this yet, and all slides i saw so far, only mentioned 4,0boost speed, not that they had a 4,2ghz XFR boost as well. 

I'm pretty sure Linus mentioned it in his video

[tom_w141]

4 minutes ago, DXMember said:

Lucky_n00b (Indonesia) Pushes AMD Threadripper 1950X to 5,187MHz, Breaks 4K in Cinebench R15

http://hwbot.org/newsflash/4691_lucky_n00b_(indonesia)_pushes_amd_threadripper_1950x_to_5187mhz_breaks_4k_in_cinebench_r15

 

btw, AMD's boost clock of 4GHz is for four cores, and XFR also goes for the four cores

You misread that. The number of cores in brackets pertains to the previous word:

Quote

3.6GHz (all cores)

 

4GHz (four cores)

 

and an Xtended Frequency Range of 4.2GHz.

 

[vanished]

3 minutes ago, DXMember said:

Well Intel also has idle clock, base clock, boost clock, tubo boost clock and tubo boost max clock

exactly, my point is it's a mess.  The days of being able to represent a CPUs speed with one number are long gone, but somehow the system we ended up with is the worst of both worlds, where we're using multiple numbers, but not really getting any additional information from them.  What they should be doing is just specifying the maximum speed x number of cores will run at.  no one cares if it's idle, or base, or boost 1, 2, and 3, just what will it get to under ideal conditions - that's it.  

[DXMember]

5 minutes ago, tom_w141 said:

You misread that. The number of cores in brackets pertains to the previous word:

 

Lisa said XFR does the same as boost if you gotz dem cooling m8

and it's in line with what XFR does on Ryzen already (Ryzen boost on two cores and XFRs on two cores) - it's boosting the boost clock past standart boost clock

[AnonymousGuy]

8 minutes ago, Ryan_Vickers said:

exactly, my point is it's a mess.  The days of being able to represent a CPUs speed with one number are long gone, but somehow the system we ended up with is the worst of both worlds, where we're using multiple numbers, but not really getting any additional information from them.  What they should be doing is just specifying the maximum speed x number of cores will run at.  no one cares if it's idle, or base, or boost 1, 2, and 3, just what will it get to under ideal conditions - that's it.  

Intel does in the spec sheets with a table for 1-10 core turbo boost 2.0 frequencies and the 1-2 core turbo boost 3.0 frequency.

 

For the 7900X, 5-10 cores active is 4.0, 3-4 cores is 4.1, 1-2 cores is 4.3.  Turbo Boost 3.0 has some dependency on TDP and temperature and shit and is 4.5 for 1-2 cores.

[vanished]

Just now, AnonymousGuy said:

Intel does in the spec sheets with a table for 1-10 core turbo boost 2.0 frequencies and the 1-2 core turbo boost 3.0 frequency.

 

For the 7900X, 5-10 cores active is 4.0, 3-4 cores is 4.1, 1-2 cores is 4.3.

I know and I'm glad they do since otherwise you'd never know what they are  

My issue is they specify a bunch of other numbers that really don't even need to exist.  They set a "base clock" and then say if there's room it will turbo up, but what if there's so little headroom it can't even hit the base clock?  Isn't that what throttling will do?  Or will it just burn itself to death to give you the "guaranteed minimum"?    If it will throttle itself under any conditions to the fastest it can safely go, then there really is no reason for a separate base and turbo/boost frequency, it's just marketing since "turbo" and "boost" sound good

[AnonymousGuy]

1 minute ago, Ryan_Vickers said:

I know and I'm glad they do since otherwise you'd never know what they are  

My issue is they specify a bunch of other numbers that really don't even need to exist.  They set a "base clock" and then say if there's room it will turbo up, but what if there's so little headroom it can't even hit the base clock?  Isn't that what throttling will do?  Or will it just burn itself to death to give you the "guaranteed minimum"?    If it will throttle itself under any conditions to the fastest it can safely go, then there really is no reason for a separate base and turbo/boost frequency, it's just marketing since "turbo" and "boost" sound good

Base clock is supposed to be the guaranteed minimum frequency under any workload, assuming you actually have a cooler rated for the TDP.  I don't know of a situation where the base clock won't be given except in shit like passively cooled mobile chips where they can throttle.

[tom_w141]

8 minutes ago, DXMember said:

Lisa said XFR does the same as boost if you gotz dem cooling m8

and it's in line with what XFR does on Ryzen already (Ryzen boost on two cores and XFRs on two cores) - it's boosting the boost clock past standart boost clock

Its 2 cores.

 

Note i'm a ryzen user so may know what I'm talking about. Now TR might be different but I doubt it.

[vanished]

Just now, AnonymousGuy said:

Base clock is supposed to be the guaranteed minimum frequency under any workload, assuming you actually have a cooler rated for the TDP.

but what if you don't?  I assume throttling will hold it back.  so what's the difference between that and not reaching max turbo?  It seems like the "base clock" is a pointless number just thrown in to create the illusion of a dividing line for no reason other than to allow them to sell "turbo boost".  Ultimately, the chip will run as close as it can to the max turbo number, unless it's held back by thermals - what changes between the base clock zone and the turbo zone?  Nothing as far as I can tell.

[DXMember]

2 minutes ago, Ryan_Vickers said:

I know and I'm glad they do since otherwise you'd never know what they are  

My issue is they specify a bunch of other numbers that really don't even need to exist.  They set a "base clock" and then say if there's room it will turbo up, but what if there's so little headroom it can't even hit the base clock?  Isn't that what throttling will do?  Or will it just burn itself to death to give you the "guaranteed minimum"?    If it will throttle itself under any conditions to the fastest it can safely go, then there really is no reason for a separate base and turbo/boost frequency, it's just marketing since "turbo" and "boost" sound good

that and also if you're browsing youtube then it probably won't go above base on a couple of cores and stay idle on most of cores, because power saving and stuff

stuff like ASUS BIOS often ignore the Intel table for boost and just boosts all cores to "boost 2.0 clocks"

if you're a power user you'll do what ASUS BIOS does and disable the C-states and P-states

[DXMember]

2 minutes ago, tom_w141 said:

Its 2 cores.

 

Note i'm a ryzen user so may know what I'm talking about. Now TR might be different but I doubt it.

Yes Ryzen 7 chips do both boost and XFR at two cores

Ryzen Threadripper is going to do both boost and XFR on four cores