Four Core Core i7-1165G7 Tiger Lake Beats Eigth core Ryzen 7 4700U Renoir in Geekbench

I like how much cache has this willow cove. Finally a bump in l1 data cache since long time and also a hefty bump in l2 cache.
That could justify some of the perf increase.

 

Intel probably ramping clockspeed too with boost clocks.

 

It gets hotter than than the sun for a few seconds just enough to win in G(c)eek(rap)bench.
Kappa.

 

Being laptop chip, CPU performance number doesn't matter the most, the whole package is much more important. Sadly for AMD, Intel, being more widely used by OEM, has still a big advantage here.
This aside, any laptop CPU number, without knowing this particular SKU chassis design (thickness, weight), the package temperature/performance sustainability and the actual battery drain is about as useful as just looking at tflops for comparing GPU.

 

Guru3d comments 2020....

 

A unreleased CPU using a new node beats an already released one by 30% on single-thread? That is good, but not super surprising.

We have come to expect rehashes of CPUs for what, almost 5 years now? I'd say that's the least they should be doing, and about time.

Meanwhile, AMD is also cooking Zen 3, which will probably be released before tiger lake and probably offer more cores, while supposedly having 15% better IPC than current Zen 2.

So yeah, time will tell if these numbers are even good.

 

Butthurt is strong with this one.

 

15w-25w zen 3 isn't coming out for some time though, obviously the desktop ones will beat this at double the TDP.

I'm more curious to see how RDNA performs on an SoC honestly.

 

Remember, the bigger the cache, the slower it reads and writes. A bigger L1 cache is only a good idea if your instructions don't fit. To my understanding) HT allows for 2 separate threads to run on the same core (and therefore share the same L1 cache), so it's tricky for Intel to find the sweet spot. Well, thanks to all the security vulnerabilities regarding HT, I assume they increased the L1 size to address some of the performance losses due to mitigations, so perhaps the L1 is meant to be more split between each thread.

 

Too bad Intel just said that benchmarks don´t matter anymore...

 

On 2 different versions of geekbench also?

 

Yeah. We also don't know TDP's which for mobile parts is huge. I wish this site would ignore these types of things with laptop/tablet chips.

 

Doesn't 10900K score higher in ST?

 

Anyone happen to notice the clock speeds?

 

what is showing there is just the nominal clockspeed, the benchmark for sure ran higher.

 

If you add .gb4 to the end of the URL (and .gb5 for Geekbench5) you can see a lot more detail. According to that it was running at 4.7GHZ which is very impressive since the entire problem with Ice lake before was that it wasn't scaling to high clock rates (they got a huge IPC boost out of it but lower clock rates, leading to basically equal performance to 14nm++++++++).

https://browser.geekbench.com/v4/cpu/15604157.gb4

 

2ghz vs 2.8ghz with no mention of the actual boosts the happening. is anyone actual NOT surprised intel won out there?

 

one of them could be fanless for all we know