Thursday, August 19th 2021

Intel's Secret Sauce at Catching Up with AMD Core Count is the Gracemont E-core and its Mind-boggling Perf/Watt

When early benchmarks of the Core i9-12900K "Alder Lake-S" processor showing performance comparable to AMD's top 16-core Ryzen 9 5950X surfaced, we knew something was up. 8 Intel P-cores and 8 E-cores, are able to match 16 "Zen 3" cores that are all performance cores. Apparently Intel is able to turn its P-core deficit around by taking a wacky approach. First, the 8 "Golden Cove" P-cores themselves offer significantly higher IPC than "Zen 3." Second, the 8 "Gracemont" E-cores aren't as "slow" as conventional wisdom would suggest.

Intel in its Architecture Day presentation put out some astounding numbers that help support how 8 big + 8 little cores are able to perform in the league of 16 AMD big cores. Apparently, on "Alder Lake-S," the 8 "Gracemont" E-cores enjoy a lavish power budget, and are able to strike an incredible performance/Watt sweet-spot. Intel claims that the "Gracemont" E-core offers 40% higher performance at ISO power than a "Skylake" core (Intel's workhorse P-core for desktops until as recently as 2020); which means it consumes 40% less power at comparable performance.
A "Gracemont" core hence doesn't end up too far behind "Skylake." The combination of high-IPC P-cores and "fairly fast" E-cores are hence able to attain performance levels comparable to 16 "Zen 3" cores. There are some limitations, though. For starters, "Gracemont" cores don't support HyperThreading, unlike "Skylake," and have a reduced ISA instruction-set compared to the P-cores.
Add your own comment

49 Comments on Intel's Secret Sauce at Catching Up with AMD Core Count is the Gracemont E-core and its Mind-boggling Perf/Watt

#26
thegnome
Looks to me like those Gracemont cores (all 8) could perform similarly to a 9700T while consuming much less power, and let's not forget the 8 other much bigger and badder cores.
Posted on Reply
#27
InVasMani
TheoneandonlyMrKThese PR piece's are getting stupid, if Intel are aiming they're next generation at against a part that's about to get a +25%(zen3+cache) performance boost near EOL and not even the competition they will face,(Zen4), I can't see the win they can.

For me it's got too damn many little cores, I mean who needs 16 small better than Skylake core's to run the backend and wants just 8 big cores for in focus gaming?!.

Surely 4/8 small OS cores and 16 Big cores is what I might have bought into not the vice versa version they're peddling.
The data actually suggests Intel should've actually probably have contrary to most people's belief gone with more of the little cores and less of the larger cores. They are much better on efficiency and occupy less space. Games don't heavily load much more than 2-4 cores anyway.
Posted on Reply
#28
Hyderz
can't wait for them benchmarks
Posted on Reply
#29
watzupken
DeathtoGnomesTake this with a grain of salt, wait for reviews.
Yup, that is correct. The thing is, Intel is pitting Alder Lake against last year's Zen 3. Matching performance in multi threading is a good feat, but may not be enough to fend off competition in my opinion. Not to mention the supposed improve comes at a great cost to power consumption as compared to the Zen 3 if rumours of the PL2 and PL4 is true.
Posted on Reply
#30
TheoneandonlyMrK
InVasManiThe data actually suggests Intel should've actually probably have contrary to most people's belief gone with more of the little cores and less of the larger cores. They are much better on efficiency and occupy less space. Games don't heavily load much more than 2-4 cores anyway.
You mentioned games, did I, I do not usually game on my main rig at all anymore in fact yet it's always on and loaded up with work, time will tell too, not all work loads or game's are fine with 4 core's, civ6 for example.

Do I agree, no.
Posted on Reply
#32
HD64G
watzupkenYup, that is correct. The thing is, Intel is pitting Alder Lake against last year's Zen 3. Matching performance in multi threading is a good feat, but may not be enough to fend off competition in my opinion. Not to mention the supposed improve comes at a great cost to power consumption as compared to the Zen 3 if rumours of the PL2 and PL4 is true.
Indeed, comparing a 250W CPU to a 140W one while talking about efficiency is somewhat dishonest.
Posted on Reply
#33
Melvis
Seriously? That first sentence couldnt be more full of bull dust if you tried.......Showing comparable performance? its a Fecking made up statement by some dude on twitter.......come on btarunr stop with the BS crap please making the artical look better then it actually is and spreading FUD......
Posted on Reply
#34
InVasMani
TheoneandonlyMrKYou mentioned games, did I, I do not usually game on my main rig at all anymore in fact yet it's always on and loaded up with work, time will tell too, not all work loads or game's are fine with 4 core's, civ6 for example.

Do I agree, no.
You literally did mention gaming regardless I feel even upwards of 4 fewer big cores in exchange for more of the little clusters has the potential to more than offset the drawbacks enough that it could've possibly added more upside. These to me look like they could offer bang for buck value over a graveyard of big cores that didn't make the cut on yields per wafer.
Posted on Reply
#35
ThrashZone
TheinsanegamerNIf the gracemont cores are so amazingly efficient and perform this well why on earth does intel even bother with the "big" cores anyway? Just make the gracemont cores the main cores and give them more cache so they can keep up with AMD.
Hi,
There have always been good cores that used less power and bad cores that needed a lot of power for the same frequency even after binning

Only option was to turn off hyper threading on all, not just the bad cores so this sounds like intel is doing this on bad cores saying they are now small cores
Only option users had prior was to lower frequency on the bad cores.

Some benchmarks disabling hyper threading actually increases scores just like it may help games perform better.
Posted on Reply
#36
TheoneandonlyMrK
InVasManiYou literally did mention gaming regardless I feel even upwards of 4 fewer big cores in exchange for more of the little clusters has the potential to more than offset the drawbacks enough that it could've possibly added more upside. These to me look like they could offer bang for buck value over a graveyard of big cores that didn't make the cut on yields per wafer.
Fair enough tangentially I did mention gaming, I'd forgot.

Over the last 20 year's we had, I used big and small cores, I didn't build a dislike for small, barely capable cores without effort.

They will suit some better than traditional CPU, but not me.

Also consider that they're competition manage to make multi core big core ccds without a graveyard of dead core's.
And little cores without HT are down 30% from the off then are smaller and less capable, more wouldn't equal more capability in most cases, and how much back end crap do people need running OS's, do still suspend useless processes you know.
Posted on Reply
#37
medi01
I don't want to imagine how licensing on this kind of things even looks like.

What is the freaking point anyway?
There are gazillion of cores per CPU, don't need that much?
Lower clocks.
Still don't need that much?
Lower clocks further.
Not happy yet?
Shut down cores.

Why bother with the "little" bazingas?
If I want my code to run on little bazingas, why don't i buy CPU with 128 bazinga's in it instead?
Posted on Reply
#38
JoniISkandar
2017 Ryzen released

Windows : what the hell is this CCX CCD, whatever just run high demand software on SMT, until they fixed scheduler...... Then intel making this big.LITTLE cpu, windows be like : ah shet here we go again
Posted on Reply
#39
Makaveli
mtcn77Jim Keller is a corporate now - what else were we expecting? The guy is the systems engineer. He will definitely trim Intel's architecture until he can have what he wants: more power!
PS: I have full confidence in him turning the power grab away from the marketing team(Product Assurance HR). Where in the world does stupid HR people know better than the engineers?
Jim Keller still works for intel?
Posted on Reply
#40
Punkenjoy
I am sure that those E core can have a good IPC, the thing that i am not sure is if they can have a high frequency.

Nobody stated the finally frequency of both cores, just the IPC. And people are so confused about IPC that they think it means the total performance of a cores.

we will see.
Posted on Reply
#41
R0H1T
Dr. DroMy concern with all of these changes is that Windows is too much of a dinosaur to be able to make use of all this tech.
It isn't Windows that's the Dinosaur, it's because such mixing/matching has a real life overhead which is hard to overcome ~ this is why the ISA & instruction sets are similar in big.LITTLE implementations or even Apple's Axx chips.
Punkenjoythe thing that i am not sure is if they can have a high frequency.
They can't, check the graph seems like they put power on X axis & how steeply power increases with performance i.e. it's much more inefficient at higher clocks & voltages than SKL.
Posted on Reply
#42
mechtech
Chrispy_Tremont/Gracemont architecture really ain't bad at all but the 40% power or performance comparison is BS because HT provides 30-40% improved numbers for Skylake and yet these comparisons are horribly handicapping Skylake to 1C1T

Gracemont doesn't need marketing lies and deception, it's a frickin' amazing bit of ultra-dense computing power that will never beat Skylake in a fair fight and isn't supposed to. It's supposed to run all the background crap at low power consumption and free up the Raptor Cove cores to focus on the threads that need maximum performance. In my experience of Denverton servers, those Tremont cores feel very much like Core2 era processing power - except today they're running 16 cores at 1.5W per core, rather than the Core2 at 65W for just a couple of cores.

Seeing a rack stacked with half-U Supermicro Atom 16C/128GB mini servers each with dual 10GbE connectivity is bonkers. A single 42U rack holds enough compute, RAM, and network bandwidth to host VDI workstations for ~1000 people and yet the whole rack is short-depth and uses barely 2KW. If that doesn't impress you then I guess you're just not that into servers.
Lol I’m not in IT or enterprise IT but sounds cool.

I did assist with hardware for our new conditioning monitoring servers. Ten 960GB ssds in a Raid5 or something. They were not cheap lol
Posted on Reply
#43
Chrispy_
mechtechLol I’m not in IT or enterprise IT but sounds cool.

I did assist with hardware for our new conditioning monitoring servers. Ten 960GB ssds in a Raid5 or something. They were not cheap lol
So much of the cost of enterprise stuff goes into validation, testing, and support contracts. It's kind of scary how expensive some enterprise stuff is and there's always the temptation to gamble on extra-redundant consumer grade stuff; If it breaks you can afford to replace it x times before it's even hit the cost of one enterprise variant - but that only works as long as there's no downtime....
Posted on Reply
#44
mechtech
Chrispy_So much of the cost of enterprise stuff goes into validation, testing, and support contracts. It's kind of scary how expensive some enterprise stuff is and there's always the temptation to gamble on extra-redundant consumer grade stuff; If it breaks you can afford to replace it x times before it's even hit the cost of one enterprise variant - but that only works as long as there's no downtime....
Indeed. Weird thing was even the Intel enterprise ssds that went in it were still 1/3 price vs the rebadged Lenovo ones. Like you said those “validation stickers” are more expensive than the hardware itself by 3 fold a lot of times.
Posted on Reply
#45
zlobby
P4-630Is a TPM in these CPU's enough for windows 11 hardware requirements or do you need to look for specific TPM motherboards as well?
It will be top kek if those CPU don't meet Win 11's security requirements...
Chrispy_If that doesn't impress you then I guess you're just not that into servers.
Oh, I'm very much into servers. Too bad every data center nowadays has CCTV and I can no longer have my way with the servers without some creepy voyeur getting horny, or having to explain myself to C-level execs. :D
Posted on Reply
#46
Dr. Dro
zlobbyIt will be top kek if those CPU don't meet Win 11's security requirements...
They do, a discrete TPM is not required. Motherboards that don't have headers for them have been passing WHQL requirements after a BIOS update. Processors that exceed the security standards include Zen 2, Coffee Lake and newer.
Posted on Reply
#47
mechtech
MakaveliJim Keller still works for intel?
Ya I read that and was like, I don't think so Tim.

from the wiki

"
In April 2018, Keller joined Intel, where he served as Senior Vice President.[21][22][23] He resigned from Intel June 2020 officially citing personal reasons.[24] Though later it was reported that he left over a dispute on whether the company should outsource more of its production.[25]

Jim Keller joined Tenstorrent as CTO in December 2020.[26]"
Posted on Reply
#48
tekjunkie28
mtcn77Jim Keller is a corporate now - what else were we expecting? The guy is the systems engineer. He will definitely trim Intel's architecture until he can have what he wants: more power!
PS: I have full confidence in him turning the power grab away from the marketing team(Product Assurance HR). Where in the world does stupid HR people know better than the engineers?
He left Intel over a year ago...
Posted on Reply
#49
Arcdar
Chrispy_Tremont/Gracemont architecture really ain't bad at all but the 40% power or performance comparison is BS because HT provides 30-40% improved numbers for Skylake and yet these comparisons are horribly handicapping Skylake to 1C1T

Gracemont doesn't need marketing lies and deception, it's a frickin' amazing bit of ultra-dense computing power that will never beat Skylake in a fair fight and isn't supposed to. It's supposed to run all the background crap at low power consumption and free up the Raptor Cove cores to focus on the threads that need maximum performance. In my experience of Denverton servers, those Tremont cores feel very much like Core2 era processing power - except today they're running 16 cores at 1.5W per core, rather than the Core2 at 65W for just a couple of cores.

Seeing a rack stacked with half-U Supermicro Atom 16C/128GB mini servers each with dual 10GbE connectivity is bonkers. A single 42U rack holds enough compute, RAM, and network bandwidth to host VDI workstations for ~1000 people and yet the whole rack is short-depth and uses barely 2KW. If that doesn't impress you then I guess you're just not that into servers.
Well said. Of course it's sad that the IPC gains since x79/x99 in gaming are ... neglectable in most cases and that we could be in a different place (than 14+++++++++++++++++++++++++++ Intel), which was proven by AMD with quite some innovation (even though they also stopped right where they had the lead and didn't go "all out" - just "as much as needed" and then always a step further....) which would be great - but it doesn't mean that not a lot changed and that especially this isn't something really amazing.

It's not what gamers were hoping for. Not even the industry. But it is a step in the right direction again and the next generations will change a lot for all (gamers and "professional users" both).
Posted on Reply
Add your own comment
Dec 18th, 2024 08:56 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts