News TSMC Celebrates Making Over a Billion 7nm Chips

[Admin]

A new post on TSMC’s website pulls back the curtain a bit about the company’s 7nm process.

TSMC Celebrates Making Over a Billion 7nm Chips : Read more

 

[spongiemaster]

A new post on TSMC’s website pulls back the curtain a bit about the company’s 7nm process.

TSMC Celebrates Making Over a Billion 7nm Chips : Read more

In related news. Intel's process engineering team has announced they have managed to successfully boil water in a pot.

 

[linuxdude]

the article is not very well researched ...

| Intel announced last month that it is delaying its 7nm chips
| “until late 2022 or early 2023.”

TSMC 7nm ~= Intel 10nm

| Part of what allowed TSMC to hit this goal was the EUV lithography
| technology it introduced this generation.

Most of those chips most likely have been manufacturered without EUV. N7 and N7P do not use EUV even on a single layer.
N7+ uses EUV, however it was introduced a lot later and still (to my knowledge) is not used for very high volume products (snapdragon 865, AMD Zen2, and so on are all N7 or N7P).

 

[samopa]

TSMC 7nm ~= Intel 10nm

How come 7nm equals 10 nm ? Please elaborate

 

[TerryLaze]

In related news. Intel's process engineering team has announced they have managed to successfully boil water in a pot.

Meh TSMCs node is limited to 3.9Ghz all core before needing 1.5Vcore and water cooling, to get up to the 10900k stock all core of 4.9Ghz ryzen needs liquid nitrogen,intel wouldn't even bother to release such a crappy chip...which is what they are doing.

 

[sstanic]

How come 7nm equals 10 nm ? Please elaborate

Not mathematically but in terms of process node names. I'm not particularly well informed, but it goes something like this: those names have become total marketing bs. Intel named theirs 10nm so Samsung said ours will be 8nm, and TSMC then said well our has to sound better so we'll name it 7nm, although they are roughly the same density. There are several relevant ways to compare process nodes, number of same-type transistors per mm2 for example. So TSMC's N7 node is roughly equivalent to what Intel calls their 10nm node. When Samsung went to develop theirs, they named it 8nm, and then improved it, and again, so now their 3rd iteration of "8nm" is called 5nm, although it is still roughly comparable to Intel's 10nm node. Marketing..

Feel free to correct me, I really don't follow it closely to know all the details.

 

[watzupken]

Most of those chips most likely have been manufacturered without EUV. N7 and N7P do not use EUV even on a single layer.
N7+ uses EUV, however it was introduced a lot later and still (to my knowledge) is not used for very high volume products (snapdragon 865, AMD Zen2, and so on are all N7 or N7P).

I think you may have limited this N7 tech to just AMD. AMD chips are still produced using 7nm DUV if I am not mistaken. The fact is that ARM SOCs are the primary driver for the billion chip milestone. While AMD is selling very well, I don't believe they contributed to this in a significant way as compared to the likes of Apple, Qualcomm and Huawei.

In my opinion, the number of 7nm chips produced by DUV and EUV should be quite close.

 

[Conahl]

Meh TSMCs node is limited to 3.9Ghz all core before needing 1.5Vcore and water cooling, to get up to the 10900k stock all core of 4.9Ghz ryzen needs liquid nitrogen,intel wouldn't even bother to release such a crappy chip...which is what they are doing.

oh ? tell that to my 3900X that is currently running @ 4250 mhz, all core, according to cpuz and ryzen master, cpuz says its using 1.240v,and ryzen master says its using 1.099 volts. this is with a Noctua NH-d15, and under load, 68C on all the cores. if any cpu needs liquid nitrogen to run at those clocks, its the intel chips.

 

[drivinfast247]

Meh TSMCs node is limited to 3.9Ghz all core before needing 1.5Vcore and water cooling, to get up to the 10900k stock all core of 4.9Ghz ryzen needs liquid nitrogen,intel wouldn't even bother to release such a crappy chip...which is what they are doing.

Luckily it doesn't have to get up that high to do the same amount of work. And it does it using less power.

 

[Don Frenser]

In related news. Intel's process engineering team has announced they have managed to successfully boil water in a pot.

The pot placed on a processor of course OC-ed

 

[spongiemaster]

Meh TSMCs node is limited to 3.9Ghz all core before needing 1.5Vcore and water cooling, to get up to the 10900k stock all core of 4.9Ghz ryzen needs liquid nitrogen,intel wouldn't even bother to release such a crappy chip...which is what they are doing.

How do you know it's the process holding back the clock speeds and not the architecture?

 

[TerryLaze]

oh ? tell that to my 3900X that is currently running @ 4250 mhz, all core, according to cpuz and ryzen master, cpuz says its using 1.240v,and ryzen master says its using 1.099 volts. this is with a Noctua NH-d15, and under load, 68C on all the cores. if any cpu needs liquid nitrogen to run at those clocks, its the intel chips.

Congratulations you got me,I meant it runs 3.9 stock and reaches 4.1-4.2 before needing way too much cooling and vcore.
The 10900k runs at ~20% higher clocks at stock at 4.9 and does so with a simple air cooler,ryzen can only dream of such clocks.

Luckily it doesn't have to get up that high to do the same amount of work. And it does it using less power.

It doesn't need to get to those clocks because it does need more cores to do the same work,yeah agreed.Also I've never seen a power draw bench with the 10900k running at 3.9 all core so it's in the air if it really uses more power or not.
I mean everybody tries to compare IPC by clocking both CPUs at 4Ghz or some other low clocks but somehow when comparing power draw on intel everybody uses the highest clocks with all TDP settings turned off...

How do you know it's the process holding back the clock speeds and not the architecture?

Either way...they are stuck at low clocks.

 

[Conahl]

Congratulations you got me,I meant it runs 3.9 stock and reaches 4.1-4.2 before needing way too much cooling and vcore.

lets see, noctua nhd15 is not too much cooling, high end water would be way to much cooling, or your liquid nitrogen. and if you call 1.3 volts way to much vcore, as this is below your " claimed " v core needed for amd's cpus to run above 3.9, and now you change your story when some one refutes your claim. and BTW, the stock clock of the 10900k, is 3.7 ghz, not 4.9. as shown here : https://ark.intel.com/content/www/u...0900k-processor-20m-cache-up-to-5-30-ghz.html thats what it is able to boost to, but depending on cooling, not for long, at it will throttle.
this here : https://www.tomshardware.com/reviews/intel-core-i9-10900k-cpu-review/2 shows this cpu uses FAR MORE power then the 3900x. 328 watts for the 10900k vs 114-117 for the 3900x.

The 10900k runs at ~20% higher clocks at stock at 4.9 and does so with a simple air cooler,ryzen can only dream of such clocks.

put a " simple air cooler " on an intel chip, and it will throttle more then it will even stay at your praised 4.9 ghz.

It doesn't need to get to those clocks because it does need more cores to do the same work

and it STILL uses LESS power then the intel chips, point is ?

I mean everybody tries to compare IPC by clocking both CPUs at 4Ghz or some other low clocks but somehow when comparing power draw on intel everybody uses the highest clocks with all TDP settings turned off...

and when comparing performance, they also use max clocks of both sides, again, point is ?

Either way...they are stuck at low clocks.

just like intels 10nm cpus ?
come on terry laze, seems like you are grasping at straws to attempt to make intel look better, when they arent as good as they once were. but hey, if you want to keep defending them, thats your choice.

 

[TerryLaze]

this here : https://www.tomshardware.com/reviews/intel-core-i9-10900k-cpu-review/2 shows this cpu uses FAR MORE power then the 3900x. 328 watts for the 10900k vs 114-117 for the 3900x.

Don't link to reviews without even reading them,come on it looks like you are pulling at straws here.
Intel has 125W TDP limit for long term and 250W only for the first 56 seconds.
Anything other than that is overclocking that the reviewers consciously leave in and don't comment on on the graphs themselves to try and make intel look bad.
If you want you can even disable the 56 seconds of tau all together at which point the 10900k will use 125W at most which is only marginally higher than 117W...

Intel set its official new PL2 watermark at a 250W TDP, which is double the 125W PL1 rating, and recommends that motherboard makers keep boost activity limited to 56 second bursts (Tau).

most motherboard vendors feed the chip up to ~330W of power at stock settings, leading to hideous power consumption metrics during AVX stress tests. Feeding 330W to a stock processor on a mainstream motherboard is a bit nuts, but it enables higher all-core frequencies for longer durations, provided the motherboard and power supply can feed the chip enough current, and your cooler can extract enough heat.

 

[Conahl]

come one terryblaze, do you actually believe intels TDP ratings ? intel sets their TDP at BASE clocks. under usage, the power their cpus use, can be higher, in some cases MUCH higher, as shown here : https://www.anandtech.com/show/13544/why-intel-processors-draw-more-power-than-expected-tdp-turbo
and BTW, the title of the article on here i linked to, should of been the dead give away on how much power this cpu uses " Intel Core i9-10900K Review: Ten Cores, 5.3 GHz, and Excessive Power Draw " the key word here, is the excessive part. from the reviews verdict : but the extra performance comes at the cost of incredibly high power consumption. seems the only way for intel to get this cpu to have the performance levels it gets, is to let it take all the power it needs, " To find the power limit associated with our chip paired with the Gigabyte Aorus Z490 Master motherboard, we ran a few Prime95 tests with AVX enabled (small FFT). During those tests, we recorded up to 332W of power consumption when paired with either the Corsair H115i 280mm AIO watercooler or a Noctua NH-D15S air cooler. Yes, that's with the processor configured at stock settings. For perspective, our 18-core Core i9-10980XE drew 'only' 256W during an identical Prime95 test. " thats at STOCK. the fact that intel allows motherboad venders to feed the chip up ~330w is the only way intel keeps its performance advantage, where it still is able to get it.

but believe what you want

 

[spongiemaster]

Yes, that's with the processor configured at stock settings. For perspective, our 18-core Core i9-10980XE drew 'only' 256W during an identical Prime95 test. " thats at STOCK. the fact that intel allows motherboad venders to feed the chip up ~330w is the only way intel keeps its performance advantage, where it still is able to get it.

There's no such thing as stock settings for an Intel CPU. Intel publishes guidelines, but motherboard makers have no obligation to abide by them. Especially for Z series motherboards, Intel's TDP guidelines are almost never followed properly within the shipped default settings, which lead to blown out power numbers like the useless, never going to be duplicated in real world applications, prime95 results you posted. Motherboard makers don't care how much power the CPU is using, they just want their motherboard to perform the best in comparison reviews which means removing the TDP limits, even when using "stock" settings for the motherboard. A motherboard that properly adheres to Intel's TDP guidelines, a few do out of the box and most if not all can be forced to follow them by adjusting BIOS settings, will not draw 300W+.

but believe what you want

 

[TerryLaze]

come one terryblaze, do you actually believe intels TDP ratings ? intel sets their TDP at BASE clocks. under usage, the power their cpus use, can be higher, in some cases MUCH higher, as shown here : https://www.anandtech.com/show/13544/why-intel-processors-draw-more-power-than-expected-tdp-turbo
and BTW, the title of the article on here i linked to, should of been the dead give away on how much power this cpu uses " Intel Core i9-10900K Review: Ten Cores, 5.3 GHz, and Excessive Power Draw " the key word here, is the excessive part. from the reviews verdict : but the extra performance comes at the cost of incredibly high power consumption. seems the only way for intel to get this cpu to have the performance levels it gets, is to let it take all the power it needs, " To find the power limit associated with our chip paired with the Gigabyte Aorus Z490 Master motherboard, we ran a few Prime95 tests with AVX enabled (small FFT). During those tests, we recorded up to 332W of power consumption when paired with either the Corsair H115i 280mm AIO watercooler or a Noctua NH-D15S air cooler. Yes, that's with the processor configured at stock settings. For perspective, our 18-core Core i9-10980XE drew 'only' 256W during an identical Prime95 test. " thats at STOCK. the fact that intel allows motherboad venders to feed the chip up ~330w is the only way intel keeps its performance advantage, where it still is able to get it.

but believe what you want

What is it with you linking things without reading them just because you like the title...?
And in this case you are even linking one review but quoting a different one, those quotes are not found on the review you linked.

These are though.

When the system has a substantial workload applied for a length of time, in this case ‘tau’ seconds, the firmware should immediately invoke PL1 as the new power limit. The turbo tables no longer apply, as those are PL2 only.

If the workload applied results in power consumption levels above PL1, then the frequency and voltages are adjusted such that the overall power consumption of the chip is within the PL1 value. This means that the whole processor reduces in frequency from its PL2 state to its PL1 state for the duration of the workload. This means that temperatures on the processor should decrease, increasing the longevity of the processor.

However, this is where it gets really stupid: the motherboard vendors got involved, because PL1, PL2 and Tau are configurable in firmware. For example in the graph above, we can set PL2 to an unlimited value, and then PL1 to 165W and 95W respectively.

the fact that intel allows motherboad venders to feed the chip up ~330w is the only way intel keeps its performance advantage, where it still is able to get it.

Again from the article that you linked to, even though you did it by mistake.
Even though this is not measured on a 10900k it shows how much more power draw there can be with minimal to realistically no improvement in performance.

 

[King_V]

Either way...they are stuck at low clocks.

Right, because clocks are all that matter. So, clearly, the FX-9590 is superior to Ryzen, by your narrow perspective?

At least the FX's thermals would be similar to Intel.

 

[TerryLaze]

Right, because clocks are all that matter. So, clearly, the FX-9590 is superior to Ryzen, by your narrow perspective?

At least the FX's thermals would be similar to Intel.

Wow, how much mental yoga do you have to do to come up with that?
Clocks aren't all that matter but they are half of all that matter,it's instructions and it is per clock also, it's a game of whoever has a higher balance between the two of them.
If ryzen could run at 5Ghz all core it would have really whooped intel instead of intel being able to counter anything amd has to offer with a 5 year old arch/process.

 

[King_V]

You chose your particular phrasing, not me.

Yes, and we could also get into the thermals which you say are the fault of the motherboards not using the power levels properly, but then that excessive power is what's used to show results, and I don't see how you think obeying the TDP limits, thus requiring lower clocks and voltages, will allow Intel to maintain its clock speed and performance advantage.

 

[TerryLaze]

You chose your particular phrasing, not me.

And there is nothing wrong with it, be it the arch or the process ryzen has more instructions but is stuck at much lower clocks which is why they can't get ahead of intel,just as intel is still stuck with less instructions but has higher clocks.
"How do you know it's the process holding back the clock speeds and not the architecture?

Either way...they are stuck at low clocks. "

Yes, and we could also get into the thermals which you say are the fault of the motherboards not using the power levels properly, but then that excessive power is what's used to show results, and I don't see how you think obeying the TDP limits, thus requiring lower clocks and voltages, will allow Intel to maintain its clock speed and performance advantage.

Not even,unlocking TDP doesn't change the max clocks that the CPUs reach,they will still stop at the stock all core clocks and over sideclocking them to this all core turbo clock is no problem and doesn't even need more TDP, not having to boost to the max single core clock might even reduce overall TDP slightly.
The only thing it does is to allow software to draw as much power as they want even if it is for no reason,which is the definition of a power virus.

 

[King_V]

So, what then? The motherboard makers are all conspiring to make Intel's CPUs appear to be less power efficient than they really are? They are drawing extra power, pushing their VRMs, and producing more heat, for "no reason"?

 

[Conahl]

spongiemaster, i think the reference to stock settings, is just putting the cpu in the board, and not changing anything regarding it, and it wasnt me that said stock settings, it was the reviewer. intels TDP ratings are based on base frequencies, as the link to AT shows, and can, and do use, more then that. i think AT has done a few articles on it, and talk about it quite often in their reviews.

What is it with you linking things without reading them just because you like the title...?
And in this case you are even linking one review but quoting a different one, those quotes are not found on the review you linked.

Again from the article that you linked to, even though you did it by mistake.

the link i posted. was to show you that intels TDP ratings, are basically meaningless, i guess you didnt bother to read it, did you? yet you accuse me of not reading? yea ok, sure. unlike you, i do read most of the articles that interest me, you on the other hand, obviously do not, cause as i said, the title for the article on here, the DEAD give away that this cpu using lots of power is the word " excessive " but i guess you didn't read that too

the other quotes were from the article on here, im sorry i didnt make that clear enough for you

TerryLaze, if clocks arent all that matter, then why is it the main thing that people that are bias towards intel, always talk about that 1st ? face it, in order for intel to still be able to claim they have the performance advantage, they are allowing this cpu to use as much power as it needs, all intel has left to compete at the moment is the clock speed advantage.