End User wrote:cphite wrote:Captain Ned wrote:Those who would give up essential fidelity to gain temporary freedom deserve neither fidelity nor freedom.
+10
-11
Infinity+1
Personal computing discussed
Moderators: renee, Flying Fox, morphine
End User wrote:cphite wrote:Captain Ned wrote:Those who would give up essential fidelity to gain temporary freedom deserve neither fidelity nor freedom.
+10
-11
End User wrote:cphite wrote:Captain Ned wrote:Those who would give up essential fidelity to gain temporary freedom deserve neither fidelity nor freedom.
+10
-11
End User wrote:ptsant wrote:instead of the ridiculous Geekbench joke that makes an iPhone look faster than a 28-core Xeon.
Link to that result.
Redocbew wrote:End User wrote:ptsant wrote:instead of the ridiculous Geekbench joke that makes an iPhone look faster than a 28-core Xeon.
Link to that result.
You would rather Geekbench be known as a benchmark which is only slightly terrible?
Redocbew wrote:End User wrote:ptsant wrote:instead of the ridiculous Geekbench joke that makes an iPhone look faster than a 28-core Xeon.
Link to that result.
You would rather Geekbench be known as a benchmark which is only slightly terrible?
End User wrote:Redocbew wrote:End User wrote:Link to that result.
You would rather Geekbench be known as a benchmark which is only slightly terrible?
Forget about Geekbench. ARM has been knocking on Intel's door for a while now. Once the mobile constraints are lifted ARM is going to be a monster.
chuckula wrote:End User wrote:Redocbew wrote:
You would rather Geekbench be known as a benchmark which is only slightly terrible?
Forget about Geekbench. ARM has been knocking on Intel's door for a while now. Once the mobile constraints are lifted ARM is going to be a monster.
Meanwhile, on planet Earth: https://www.phoronix.com/scan.php?page= ... core&num=1
End User wrote:chuckula wrote:End User wrote:Forget about Geekbench. ARM has been knocking on Intel's door for a while now. Once the mobile constraints are lifted ARM is going to be a monster.
Meanwhile, on planet Earth: https://www.phoronix.com/scan.php?page= ... core&num=1
Very interesting benchmarks. Thanks for sharing.
The Cavium ThunderX SoC in those benchmarks uses 64-bit ARMv8 cores, manufactured on a 28nm process. The last Apple ARM based SoC that was on the 28nm process was the A7 back in 2013.
The current A10X (10nm) has over 3x the single core performance of the A7. Cavium states that their 2nd gen 16nm X2 product delivers 2-3X higher performance.
Very interesting indeed. The pace of progress is staggering. This bodes well for ARM in laptops and on the desktop.
chuckula wrote:Yeah, keep drinking the koolaid. As for the A10X being "over 3X the single core performance of the A7" it sure doesn't seem to show up in the complaints about iOS performance that have been coming out recently.
chuckula wrote:But let's make a friendly bet: Since ARM is so magical and great, a 2020 Apple Miracle chip should beat a 2020 Tigerton across the board in single-threaded workloads with a 1GHz clock speed deficit. As for multithread, a 4-core Apple Miracle Chip should be at least as fast as an 8 core Tigerton.
chuckula wrote:I mean, given your level of delusion, a single-core Apple magic-chip ought to be easily equivalent to an 8-core Skylake X running in multi-threaded operations, so I'm intentionally stacking the odds in Apple's favor here.
Redocbew wrote:To add on to JBI's point from earlier: while there isn't any technical reason I know of for why a chip from ARM couldn't be successful on the desktop there also isn't any reason I can think of for why one would be particularly suited to the desktop. What you're calling "ARM performance" is really just a bunch of engineers who are good at doing their jobs, and they happen to be making ARM chips for whoever needs them instead of x86 chips for AMD or Intel.
Usacomp2k3 wrote:End User wrote:Usacomp2k3 wrote:We reached "good enough" computer for most tasks, especially media consumption back in the Haswell era.
I was thinking Core 2 Duo.
The i5-2500K was a big step over the dual core C2D's. Now the C2Q like the Q6600? Maybe.
Redocbew wrote:To add on to JBI's point from earlier: while there isn't any technical reason I know of for why a chip from ARM couldn't be successful on the desktop there also isn't any reason I can think of for why one would be particularly suited to the desktop.
bthylafh wrote:Redocbew wrote:To add on to JBI's point from earlier: while there isn't any technical reason I know of for why a chip from ARM couldn't be successful on the desktop there also isn't any reason I can think of for why one would be particularly suited to the desktop.
I have a hunch that for ARM on the desktop to really shine (at least vs. Intel), you'd need a ton of cores plus workloads that can be heavily parallelized. Maybe Apple could swing that because of how integrated their stack is (and Grand Central Dispatch, to a lesser degree) and how little they care about back compatibility. I bet they could put together a decent Macbook with a big.LITTLE setup.
Redocbew wrote:To add on to JBI's point from earlier: while there isn't any technical reason I know of for why a chip from ARM couldn't be successful on the desktop there also isn't any reason I can think of for why one would be particularly suited to the desktop. What you're calling "ARM performance" is really just a bunch of engineers who are good at doing their jobs, and they happen to be making ARM chips for whoever needs them instead of x86 chips for AMD or Intel.
Glorious wrote:But the idea that they are just going to blow Intel out of the water because of the "magic of ARM" is insane
End User wrote:Performance on par with Intel is the only thing Apple needs to accomplish initially.
End User wrote:There are strong hints that Apple has already achieved this with its mobile ARM variants (i carry examples of that around with me every day).
End User wrote:As I said in another post, Windows 10 on ARM (the native code portion of that product) runs well on a previous gen mobile ARM SoC. The writing is on the wall.
Glorious wrote:Apple can choose whatever hardware it likes, as it has switched between 3 different architectures in my lifetime and could easily (for the macbook airs or something) switch to a fourth for the low-end. The vast majority of people would literally not even notice.
Glorious wrote:Saying that Apple will make an *APPLE* ARM chip for *APPLE* products means nothing when it comes to ARM generically: It won't make the innumerable pre-existing ARM processors from other vendors that perform poorly magically perform better.
Glorious wrote:ARM has simply not taken over the datacenter world at all, despite the concern about power usage and how the major vendors who buy literal truckloads of these chips could (and likely already have) ported their entire stack over to it.
Glorious wrote:But sure, back to your "hints" and conflation of low-power performance with high-power performance.
End User wrote:I'll just come out and say it. The A11 Bionic is roughly equivalent to an i3-7100.
Captain Ned wrote:End User wrote:I'll just come out and say it. The A11 Bionic is roughly equivalent to an i3-7100.
Which makes it your claim to prove.
End User wrote:Exactly. I don't give a fig about other vendors. Neither does Apple.
End User wrote:Windows 10 on ARM (the native code portion of that product) runs well on a previous gen mobile ARM SoC. The writing is on the wall.
Glorious" wrote:Saying that Apple will make an *APPLE* ARM chip for *APPLE* products means nothing when it comes to ARM generically:
End User wrote:ARM has been knocking on Intel's door for a while now. Once the mobile constraints are lifted ARM is going to be a monster.
End User wrote:I'll just come out and say it. The A11 Bionic is roughly equivalent to an i3-7100.
Glorious wrote:These chips could be, and in practice basically *are* "Apple" instruction set processors. Apple products are infamously proprietary.
chuckula wrote:End User wrote:Redocbew wrote:
You would rather Geekbench be known as a benchmark which is only slightly terrible?
Forget about Geekbench. ARM has been knocking on Intel's door for a while now. Once the mobile constraints are lifted ARM is going to be a monster.
Meanwhile, on planet Earth: https://www.phoronix.com/scan.php?page= ... core&num=1
Glorious wrote:It's not ARM at all, it's just Apple.
Glorious wrote:And it's not Apple knocking on Intel's door, no. It is simply Apple closing *THEIR* door to Intel.
Glorious wrote:I'm just rejecting your inane fanboyism, especially the part of it were you are so delusional that you plainly pretend that you didn't make the grandiose statements you JUST DID.
Glorious wrote:I'm glad you have something to believe in.
Glorious wrote:Which, again, is my point!
Glorious wrote:Apple products are infamously proprietary.
Buub wrote:I predict will simply see a convergence where arm comes up to the same level. Then it will simply be juggling performance-per-watt instead of raw performance.
Glorious wrote:Why do you even post such garbage?