Intel renames nodes: 10nm+ and 7nm become Intel 7 and 4

So, an Intel 7 and a Ryzen 7 are like basically the same thing just a different brand, like a Gigabyte 3080 and an Asus 3080. Thanks for making things clearer for the masses Intel.

 

The simple truth is that node size has been an irrelevant metric for a long time. Really it hasn't represented an actual on silicon feature in probably 20 years or so now since like 1997 ish. At that time, though, it was the opposite; the node name was actually LARGER than the physical dimension of the gate size. Fast forward to 2011 ish and it became what we see today, with the Xnm being a number that is much smaller than the actual gate length. I think the move away from pretending it is actually some physical measurement makes more sense, and it does bring more parity to those that don't necessarily understand the process as well (Intel's 10nm node is more in line with TSMC/Samsung's 7nm, their 7nm node is more in line with TSMC/Samsung's 4nm etc).

Would it be nice if there was a better metric they could all use such that the nomenclature actually is tied to a physical attribute? Sure. I'd love to see all of them transition over to a simpler "Density Metric". But even a density metric would need more clarity (Logic density? Memory density? Integration density?), and even then it's obvious that when you have a benchmark of any kind that becomes a marketing term, companies over time will find ways to game the system to inflate their perception vs. the competition.

 

is this even remotely surprising ? Even on this very forum where we are a crowd that is above average the population on how much tech savy we are ... i saw comments like " omg intel now is making 10nm work tsmc is already on 7 nm ... etc etc " and when intel was talking about their 7nm they where noting that tsmc is preparing the 5nm ...soooo reaaaaaally this is not surprising at all since it seems the naming tsmc and the rest used worked ...they adapt accordingly

 

Yeah remember that, actually had two of them myself back in the day.

Like the Athlon XP 2400+ was supposed to equal 2.4 GHz Pentium IV even if it had a lower clock frequency.
And perhaps practically it did, but surely a confusing naming scheme.

 

Officially this was a rating against the athlon thunderbird one that it was equivalent to a thunderbird clocked to 2400 at least that what amd was saying

 

Honestly, this whole nm war has became totally silly and meaningless since every company names it differently. All I care about is performance to watt ratio and maximum consumption of the final product.

 

This was the one time they could have finally switched to transistor density (something like transistors per square millimeter) but nooooo they had to go in the total opposite direction.

 

Intel is back to trying to compete with vaporware and obfuscation. *golf clap* Makes you feel sorry for the Intel apologists. Intel: "Even if we cannot make competitive products, we will change our terminology to make what we sell sound more competitive!" Now, that's a winning strategy [not.] Every day the situation becomes more idiotic. Wasn't changing the TDPs enough? Guess not.

 

Intel's quoting the numbers it wants to quote and nothing else. So, no--Intel's 10nm does not = AMD's 7nm. That's so lame, really. The truth is in the TDPs--but with Intel those are always changing...

 

Seems to me Intel has been drinking again.
We all know theres no point in arguing with a drunk.

 

Intel clown inside

 

No you remember Pentium PR rating and AMD Athlon XP 1600+? I next would be 20 core cpu, which would be in reality 4 cores + 16 very small cores, just to look beefy..

 

I'm even old enough to remember when Cyrix was naming their CPUs backthen in late '90s to match Intel counterparts: Cyrix P166+ (133MHz) to fight Intel Pentium 166MHz, Cyrix P200+ (150MHz) to match Intel Pentium 200MHz, not sure but think Cyrix even had P233+. Most old farts here know how that ended up lol... iirc I had a Cyrix P166+ for a while and that thing was a POS, couldn't OC that crap not even a bit. Not long time has passed since I've bought P166+ and iirc got me Intel 233MHz MMX few weeks later, it was night and day difference.

Year by year it's getting harder to track down and store all that info in our geek brains about what manufacturers are doing in terms of manufacturing process, all those "my dad is stronger than you dad" nm fights about who's producing smaller nodes. All of us consumers should not care that much about MHz/GHz, nm... ffs just give us more efficient CPU cores followed by lower power usage and lower temperatures, OS and SW that could make propper use of all those cores and that's all.

Not trying to start another rant or derail this topic but personally I'm not even thinking about trying to get the max out from the CPU with OC past few year, consumer OC as we knew it 5, 10, 20 years ago is history, without huge money burned on LN2, chiller or at least massive WC system there is no 30%, 40% or even higher CPU performance uplift with OC as we remember back in lets say Celeron 300A, 1st gen 1156 and 1366 i5's and i7's and, 32nm Westmere's, i7-2600K days... and almost any a bit more than average Joe could do that OC bacthen. When I told 15, 20, 25+ years old OC stories to my kids they almost think I've gone mad Lucky me I still have almost all that fully functional HW to prove my stories but they"re not interested in OC, not a bit, fully understandable.

Rant is over. Intel, AMD, TSMC, Samsung... whoever. Efficiency, performance and lower power consumption/temps is all and consumers should care about whwn chosing CPU, GPU... I don't even want to know or care about this stupid nm production fight their marketing is trying to bomb us everyday with.

 

Yeah, you're not kidding. The P166+ was the very first CPU I fried trying to overclock it. It had zero headroom. Went up in smoke instantly. I replaced it with a 166MMX and the 233MMX soon after.

 

Maybe it´s better ti stop discolosing the process being used and just sell the products base on their performance, power envelopes and prices but this is getting ridiculous...

 

The intel hatred is glaring in this thread. Intel just clarified that their 10nm is comparable to others 7nm, so they call it 7 like everyone else.... No confusion, nothing misleading. Just conforming to the standards that TSMC and Samsung are going by.

 

Yah - this has been known by most people looking at feature sizes for a while now but you have the usuals coming out in this thread.

The truth is in TDP? Really? The truth about process manufacturing is in something that each company defines on it's own and is designated for specific chip designs? That's a real stretch of logic lol

 

lol, my Cyrux P166+didn't went up in smoke but it was super-unstable even with +1MHz on the FSB, multi was locked on these iirc. Sure motherboard I was using was quite nice for OC, sadly but can't remember the brand and mobo model, was so confused trying to OC that damn thing asking myself: "...what am I doing wrong, did I get the dud chip or what!? Maybe these are OCing somewhat different?..". At least 10 different Intel and AMD CPUs OCed prior to Cyrix P166+, all of them had some OC headroom, some more, some less, but this P166+: nada! Nothing helped: higher voltage made OC attempts even more unstable. Tried using huge air cooler (custom made CNC miled aliminium block + 120mm fan attached to it I've used to OC Pentium 100MHz, or it was Pentium 120MHz, f*ck me I'm geting old, can't remember all those details, thing was weird looking, almost fugly, but cooling capacity was insane, not sure but around 1.5Kg weight, custom made brackets keeping all that mass screwed to hang from the top of the case) and zero difference. Had to give up OC that POS few days later.

Funny enough but thunk none of local OCers backthen who knewed what nm process was used for production of any of our CPUs used for OC in mid/late '90 and early 2000s, none of us knew or cared about that. Not sure but that "production node nm thingy" I've heard about maybe in Celeron 300A (Slot 1) OC days, or it was back in the AMD K7 TBred A, B, Barton OC days. No one I knew backthen was interested in that, all we cared about was "moar MHz and higher FSB speed because that's where the power was coming out!"

Good old days, cheap OC => massive performance uplift...

 

Except it's not 7nm, and it's also better than TSMC's 5nm, so you're kind of just displaying your lack of knowledge.

 

They don't need to, cause it's faster in everything, not just Cinebench .