Mobile processors have worse IPC compared to Desktop processors, at least in some of the Spec workloads, this is not only because of cache differences, but also because LPDDR4X has worse timings compared to standard DDR. So this has to be taken into account when comparing Tiger lake IPC to Zen3 Desktop.
I don't know where you get your confidence from.
Skylake 8C desktop has 17% higher IPC over Skylake 4C laptop from double the cache and a desktop memory subsystem. This should clue you into the nature of the benchmarks you are trying to extrapolate from.
Looks like 11900K QS numbers above your post are a few % higher than the 5800X on non-memory sensitive benchmarks, too.
11900k will probably just be for people who want good samples and the absolute best gaming/overall 8 core performance when overclocked. 11700k also has 8 cores and 16 threads, which will be the real killer offering from Rocket if Intel is remotely aggressive with pricing (as long as it's widely available).Honesty, I don't have a dog in this fight.
Based on Ian's numbers it seems as though Intel has between 6.9% and 14.4% to "make up" through memory subsystem and clock by moving Sunny Cove to the desktop.
If they made it up only through clocks then...
4800 x 0.069 = 331.2 on the low end of this estimate (RL at 5131MHz)
4800 x 0.144 = 691.2 on the high end (RL at 5491MHz)
Assuming they get a few percent through the transition to faster desktop memory subsystem then you are right and they are definitely competitive with 5800x.
But it still seems like the two parts will be very close in overall performance right?
Will Intel price 11900k at $450 to sway AMD buyers to Intel assuming performance is a wash?
If they go over $450 it'll be hard to justify not going with the 5900x.
If they go a bit lower like $399.99 AND have slightly better gaming performance they could be very attractive to a large part of the market.
Okay so it's not as bad for Intel as I posted above. I can admit that. But they still have the unusual situation of having the "ceiling" with the 5900x/5950x being available.
Or, best of all situations we could get a bit of a price war! Either way the release of RL will fire up the market.
5950x vs 1065 - SPEC2017 5950x 1065 SPEC2017int - 10 scores 1.514 1.323 -14.4% SPEC2017fp - 12 scores 2.413 2.335 -3.3% Combined 3.675 3.438 -6.9%
5950x vs 1065 - SPEC2006 5950x 1065 SPEC2017int - 12 scores 13.57 11.47 -18.3% SPEC2017fp - 6 scores 18.62 17.05 -9.2% Combined 22.880 19.995 -14.4%
It's amazing what competition can do...
Okay so it's not as bad for Intel as I posted above. I can admit that. But they still have the unusual situation of having the "ceiling" with the 5900x/5950x being available.
The irony here is that we are talking about product that should have originally came out in 2017, on 10nm. Yup, that is how long Intel is having process issues and they still have to release 14nm frankenstein to compete with AMD.
The other competitor at least has the benefit of doubt for being underdog and having horrible execution for a decade+. But releasing CPU in 2020, on 7nm with 32MB of L3 that is beaten by what is 2015-2016 design is not exactly what proper technical competition should bring us.
The only things competition has brought us is more core and better prices per core.
I am hopeful that Jasper Lake/ZEN4 designs will finally break this BS cycle and bring proper, 2020 era designs instead of these compromise ridden designs. X86 survival depends on it, there won't be any chances left if ARM removes server volume, while at same time AppleARMs takes away further chunk of laptop pie.
AMD had the advantage of not having to manufacture chips or mess with process node tech at all.
If TSMC had gotten stuck on 12nm, where would AMD be? If GloFlo had enforced their contract with AMD and not allowed them to split up between 7nm TSMC chiplets and 12nm IO at GloFlow - where would AMD be?
It took them two generations of 7nm TSMC parts on Zen to beat Skylake, and now they have a part that cant supply a tiny fraction of a very small part of the market.
AMD is not in charge of its own destiny, they are symbiotic with the fabs. TSMC's accomplishments are not AMDs accomplishments. Intel just had to wake up and see that they had dangerous competition and a reason to get those designs out of the lab and into the market.
AMD is not in charge of its own destiny, they are symbiotic with the fabs. TSMC's accomplishments are not AMDs accomplishments. Intel just had to wake up and see that they had dangerous competition and a reason to get those designs out of the lab and into the market.
True but that in the end turns out to be one of the faces of competition. You can stand still on a great product and find yourself obsolete in a few years in the tech sector, hopefully there's never again a 5 years standstill on architecture, doesn't benefit anyone in the world when things stagnate.
I do hope the same competition from ARM pushes x86 to new heights, can't have a mere 25% increase in single thread after 5 years (thinking a single core at 5Ghz of Skylake back in 2015 vs todays Zen or Rocket).
Mobile did see huge gains on laptops, now it's time for desktops to shine.
This is true, it's kind of a side benefit to companies like AMD though.
There is no reason Intel could not have introduced Rocket Lake in 2018 instead of the 8th gen.
Most likely AMD would not have survived that, as it would have been competing with Zen 1+ 2000 series. You'd be literally talking about 80% better single thread and 35% better multi.
Geekbench comparison :
There is no reason Intel could not have introduced Rocket Lake in 2018 instead of the 8th gen.
AMD by their own admission never expected to be in this position relative to Intel. They planned for Intel to be further along in CPU development, so it is reasonable to think that they also never expected demand for Zen would be at these levels. Their slow market share growth projections reinforce this. The fact that Intel screwed up much more than most expected seems to have caught them and TSMC by surprise. Lead times for fabs being so long also means no short term solution.AMD had the advantage of not having to manufacture chips or mess with process node tech at all.
If TSMC had gotten stuck on 12nm, where would AMD be? If GloFlo had enforced their contract with AMD and not allowed them to split up between 7nm TSMC chiplets and 12nm IO at GloFlow - where would AMD be?
It took them two generations of 7nm TSMC parts on Zen to beat Skylake, and now they have a part that cant supply a tiny fraction of a very small part of the market.
AMD is not in charge of its own destiny, they are symbiotic with the fabs. TSMC's accomplishments are not AMDs accomplishments. Intel just had to wake up and see that they had dangerous competition and a reason to get those designs out of the lab and into the market.
Yes there is. A backport like this isn't something a company can just do at the snap of their fingers. It takes a fairly significant amount of time.
Suggesting Intel could have done RKL-S - a backport of Ice Lake-S at that - the year after CNL-S was planned to launch is a little bit hopeful, to say the least.
AMD by their own admission never expected to be in this position relative to Intel. They planned for Intel to be further along in CPU development, so it is reasonable to think that they also never expected demand for Zen would be at these levels. Their slow market share growth projections reinforce this. The fact that Intel screwed up much more than most expected seems to have caught them and TSMC by surprise. Lead times for fabs being so long also means no short term solution.
In other words, your argument is exploiting hindsight to appear wise.
So then the reason Intel could not have done Rocket Lake in 2018 is because they had no plans to do it.My understanding is that ICL was designed in 2017. If they had backported to 14nm as a fallback strategy during that design, it would have been ready in 2018. Intel failed at being paranoid about 10nm, and they failed while expanding capacity in 14nm. Even if they had waited until 2019 for Rocket Lake, it would have been Zen 2 vs Rocket Lake, which would have been a disaster for AMD.
However, it's not like their bottom line suffered. 2020 was their best year ever in terms of revenue and profit. Their revenue in absolute dollars grew more than AMDs entire revenue stream, which of course is record-breaking for AMD.
Anyway, spilt milk. The impending Zen 4 / 5nm early 2022 vs Alder Lake will probably keep both companies on their toes for the next couple of years.
However, it's not like their bottom line suffered. 2020 was their best year ever in terms of revenue and profit. Their revenue in absolute dollars grew more than AMDs entire revenue stream, which of course is record-breaking for AMD.
How is this a personal attack? Saying the argument is presented as to appear being wise is about the argument not you. Don't you understand the difference?Wow really, AMD fanboys resorting to personal attacks. Who would've thunk it. Except, you're typical. Welcome to my ever increasing list of ignore recipients.
So now, we're going to see top bin 8 core with the highest 1-2 thread performance. I think these will wind up outperforming 12 and 16 core Zen 3 on games and typical lightly threaded workloads - which is to say like 99% of all workloads.
You said "your" argument, not "the" argument, so it was personal.How is this a personal attack? Saying the argument is presented as to appear being wise is about the argument not you. Don't you understand the difference?
As I was informed previously on this very thread, Intel's Cove cores have been designed in such a way so as not to be tied to any one particular process node.
Yea, I remember when Skylake came out, I remember someone from intel stating that it would be their most important architecture ever. Kind of ironic, no?This is a good insight into yet more of this new dynamic between AMD and Intel.
AMD puts their best single core clock parts in the 16 core, 12 core, and finally 8 core parts.
But for this go round Intel will be topping out at 8 cores so of course their best parts will be 8 core.
If AMD needs a little "kick in the pants" for the 5800x they can always release a 5800xt part right?
The other insight in this thread I found interesting is that we should remember it did take AMD quite a few generations to catch and finally beat the 5 year old Skylake design. Kind of shows how robust Skylake is.
We have a healthy competition here