Presumably, with such an alarming benchmark, there will be a quick followup.
Last edited:
First benchmark of Meteor Lake online
- Thread starter Fred Chen
- Start date
Presumably, with such an alarming benchmark, there will be a quick followup.
eh… Given that we have seen industry standard ARM cores running in the high 3GHz range at VLSI last year at far lower power than intel 7; I am unconcerned by an engineering sample running at the unbelievably low speed of 500MHz. Furthermore this is presumably a cut down part (so far they have shown 6+8 and 2+8 dies, no 4+8 die), on a website that has meme status for their shoddy “benchmarking”, and reported on by a “news” outlet with dubious credibility based on the BS they report about TSMC/Samsung/intel. After all, according to them intel cancelled their N3 orders and Samsung 5LPP yields were 10% for Qualcomm.
This of course says nothing about how ludicrous 500MHz is. Maybe without the heatsink or at voltages so low you can’t even reliably operate the devices. But if we do take this to be indicative of final Meteorlake this is even worse than Cannonlake (which doesn’t even seem possible).
This of course says nothing about how ludicrous 500MHz is. Maybe without the heatsink or at voltages so low you can’t even reliably operate the devices. But if we do take this to be indicative of final Meteorlake this is even worse than Cannonlake (which doesn’t even seem possible).
Shady business this benchmarking. I remember writing some benchmarking code way back when. We were well under one million instructions per second (MIPS) but our master benchmarking skills got us there. I have never trusted a benchmark since.
This happens in engineering prototypes without full thermal solutions, or is deliberately throttled back when someone is just running the benchmark for test/verification work. The chip is probably not in final stepping yet, there may even be timing issues they wanted to avoid. In summary, this data point is meaningless.
That seems more likely, thanks. Then it would not be the real or final Meteor Lake sample.
Xebec
Well-known member
User Benchmark is literally the worst benchmark as it’s either a very basic or made up numbers. Many forums actually ban or at least disallow its use as a comparison tool.
Re: 7nm, the article is using old Intel nomenclature — Intel 4 (now) was “Intel 7nm”.
I don’t have first hand knowledge but most indications seem to show Meteor Lake as fairly healthy, both in terms of IPC and ability to clock.. though it won’t be a 6 GHz miracle like Raptor Lake. Intel ‘4’ looks good and will bring Intel back to process parity on the CPU side with AMD at least for Mobile.
Re: 7nm, the article is using old Intel nomenclature — Intel 4 (now) was “Intel 7nm”.
I don’t have first hand knowledge but most indications seem to show Meteor Lake as fairly healthy, both in terms of IPC and ability to clock.. though it won’t be a 6 GHz miracle like Raptor Lake. Intel ‘4’ looks good and will bring Intel back to process parity on the CPU side with AMD at least for Mobile.
UserBenchmark is actually a large database with mostly normal numbers, so this low figure stood out. It looks most likely to be a sub par setup or sample, so not a representative.
Xebec
Well-known member
That’s fair
Just don’t take Userbenchmark too seriously
. Is your computer a “UFO, an Airplane, or a Battleship”. — with sources not identified for determining each.. makes for a fun but not very useful comparison FWIW, on the same scale of ‘could be a fact, might not be a fact’, there is a claim that Meteor Lake ‘will definitely hit 4.7 GHz’, ‘is likely to hit 5.0 GHz’, and the engineers are ‘aiming for 5.4 GHz’. At 5.0-5.4 GHz it won’t quite unseat Raptor Lake’s 6+ GHz on the desktop (refresh) but would make a whole ton of sense for laptops..