My First Legendary Score In Time Spy (7945HX/4080)

[uncut fjord]

Lenovo Legion Pro 7 Gen 8 16ARX8H
Ryzen 9 7945HX (undervolted via per core Curve Optimizer)
NVIDIA RTX 4080 (+212 core clock at -0.910mV)
Samsung 990 PRO 2+4TB SSD
Kingston FURY Impact 2x32GB RAM (KF556S40IBK2-64)

CO was set in UXTU
RAM Overclock was set in UMAF

[safe rivet]

because I am interested, please follow the proper format

[uncut fjord]

My fault, does that look better?

[terse plank]

great cpu score but gpu score is on the lower side of the rankings

[uncut fjord]

Yes, I was using my everyday gpu overclock. I think I can get away with a +800 memory in the benchmark but not it eventually crashes

[uncut fjord]

Ok this is weird. For some reason I can't reproduce the CPU score even when I'm using the same settings. I have no idea why this is

CPU score can never go past like 16.2K

I haven't tweaked anything since I got the 18K

[potent steeple]

Welcome to the 4080M big league 🙂

[terse plank]

18k is a big score for that cpu

[uncut fjord]

Yea but I cant reproduce it anymore even with the same settings and Idk why

[last tangle]

Is UXTU the same as Universal x86 tuning utility

[safe rivet]

yes

[last tangle]

I was able to post a legendary score on time spy http://www.3dmark.com/spy/60984849

Looking for advice to push further

[uncut fjord]

What settings did you use?

[last tangle]

CPU: +150MHz 10x PBO -25 all core (After updating x86 tuning utility, my per core CO got deleted somehow. It took me so much time to have a 100% stable CO and i don't have any pictures unfortunately)

Memory: 5600MHz tuned 2100FCLK . (I also have a tuned and stable 6000 profile but for daily driving the extra noise isn't really worth it for me.

I will run timespy again with the 6000 profile next vacations for sure)

GPU: 2700MHz @975mV +800 memory (Obviously not stable in most games I play, just a fast and dirty OC for highscore)

This was just a very fast and dirty OC to see how high I can go with flydigibs2 pro that I got yesterday haha

[uncut fjord]

Did you configure any voltages like vdimm, vsoc, etc?

[terse plank]

its winter open your room window

and youll get like %15 more scores

this is not cheating at all

this is just eskimooing

[quartz valve]

How is yours legendary?

I scored higher, yet mine is excellent only?

Nvm, my dumbass just realised you have a different cpu lol

Damn, my stock 14900hx is better than a tuned 7945hx according to timespy apparently???

[nimble ice]

Lol why u have msi afterburner in the background while doing a timespy?

If u're going for a record close everything the background

[nimble ice]

On ur zentimings only vsoc is listed, don't think u can tweak ur vdd and vddq, but maybe vsoc is changeable

[terse plank]

no its not, only on timespy

timespy is just a bad cpu benchmark

[last tangle]

not necessarily. All 3d mark benchmarks favour Intel CPUs for some reason.

You can confirm this by looking at GPU only benchmarks for any GPU or extreme OC for xx90 (desktop/laptop) and all of these cards will be paired with an Intel CPU mostly 13900/14900 CPUs

14900HX might be stronger than 7945HX but I don't have enough info and research about the CPU to talk about it.

[quartz valve]

I think the 14900hx is slightly better regardless

But yeah timespy cpu benchmark favours Intel for whatever reason

Cinebench more accurate

I get 32,000 cinebench r23 multicore on my stock 14900hx

I think if I tune my ram timings or undervolt it can definitely be around 36k

[terse plank]

because it cant take advantage of L3

its just a bad benchmark

that doesnt reflect real world scenarios

[uncut fjord]

Your CPU needs upwards of 190W to get the same performance mine does at 130W. But yea lets pretend yours is better because of one benchmark that's documented to favor yours as opposed to every other one as well as real world examples lmao

[uncut fjord]

Here's the benchmark scores for 7945HX on cinebench r23 from notebookcheck. Your score doesn't even beat the average LMFAO

[uncut fjord]

At less wattage mind you. Around 130-140 compared to Intel's 190

[quartz valve]

lol calm tf down, i literally don't give a shit

they are neck and neck regardless

i realised that timespy cpu benchmark was useless anyways

regardless on the same notebook check website

the 14900hx scores 2150 single core and the 7945hx 1919

if you want to split hairs, a higher single core will give you better fps in games

theyre literally the same perf, but ig you want to split hairs, but the 14900hx is technically better because it has about the same single core score as the newer 9955hx, lol.

in r23 atleast'

[last tangle]

idk what wattage this was on but my 7945HX stock does 35k in r23 and tuned with ram OC at 6000 36-37-37-34 does 37.6-37.8k

[uncut fjord]

Keep coping I guess

[quartz valve]

Not cope buddy

7945hx has a 12% lower single core score

14900hx has the same single core score as the 9955hx

[nimble ice]

Lol u guys really arguing about whose cpu has a bigger dick

[nimble ice]

OP is the one who started this shit though

[last tangle]

wait till he finds out people even OC their Northwood pentium 4 chips in this age. I have a a pentium 4 sytstem I used to OC (killed 2 and learned a lot). I also OC my old 4790K in a freezer cooled setup. Should I be sad that I can't score as high as 14900?

[nimble ice]

That goes for the both of them tho, neither of them were friendly lol