one can surely hope![]()
-
Not really a matter of hope. Its a well known fact that the physics test in firestrike is worthless. Lol
http://www.3dmark.com/3dm11/9882214
I'd have to get off my lazy butt and re-run it to get parity for Windows 10 but it shouldn't be that big of a difference. -
it is, hm? learn something new every day i guess
-
That score is typical of the broadwell cpus which shows the performance gain from haswell.
-
Broadwell has a 5% IPC improvement so there is something else going on (I'd bet on the inaccuracy of firestrike more than anything).
-
Could be but in the gt80 test and tune thread I have seen broadwell physics score in the 12k or higher range on a few machines.
Sent from my SM-G925T using Tapatalk -
My 4790K did 12349 @ 4.5GHz... Hmm... Still not bad for BGA junk...
http://www.3dmark.com/fs/6136258 -
In 3DMark11 or in firestrike? 3DMark11 is a true test... I've only managed to trip my A/C adapter with 3DMark11, not Firestrike. It's a much better test and I've had overclocks on the CPU that were stable with standard tests BSOD on the physics test of 11.
-
The thread is for firestrike and the test above I was referencing was firestrike and the performace is common for broadwell so no worries.
Sent from my SM-G925T using Tapatalkhmscott likes this. -
Yeah he got real lucky there with that chip, no doubt. The GT72 keeps it cool and he doesn't hit his power limits.
Yeah, I'm wholeheartedly convinced now: broadwell's L4 cache is helping those physics scores IMMENSELY. My 3.8GHz run is a few HUNDRED points less, at a faster speed.TomJGX, hmscott, jaybee83 and 1 other person like this. -
It would appear that is indeed the case. I did some investigating and the L4 is so useful that the 5775C @ 4.2GHz gets a 30% lead over the 4790k in PCMark! Pretty impressive.
-
Yup! It's actually quite funny because it makes broadwell superior to Skylake in a LARGE number of instances. But y'know it's expensive for intel and most users won't see the benefit so...
-
I think most users would see the benefit as long as the software leveraged it properly. The average without PCMark was 7% - that's quite a bit of a performance boost in PCMark.
-
HaloGod2012 Notebook Virtuoso
what other benchmarks shall I run for a better physics test?
here is 3dmark11: 9k physics score
http://www.3dmark.com/3dm11/10361993Last edited: Oct 3, 2015hmscott likes this. -
funny that u mention that, the highest power draw ive ever seen was actually with 3dmark11!
-
Ethrem, adding Firestrike Physics result for the GT80 SLI-263, with 5950HQ:
http://www.3dmark.com/compare/fs/6133190/fs/6031297/fs/5708173
Physics: 13681
We need a new 980 with Skylake to add to the compares
Last edited: Oct 3, 2015 -
HaloGod2012, comparing 3dmark11 with the GT80 SLI-263 with 5950HQ:
http://www.3dmark.com/compare/3dm11/10361993/3dm11/10169847
Physics: 10893
We need a new 980 with Skylake to add to the compares
-
Guys, @HTWingNut has posted first benchmarks from a 6700K, still a 980m, but at last we can compare a 6700k Firestrike Physics result
This is a first/early result, checking the result details it turbo'd to 4.2ghz, but likely there is more tuning to do to improve, so I consider this as preview.
http://www.3dmark.com/compare/fs/6133190/fs/6031297/fs/5708173/fs/6136972
6700K Physics: 12680
Here is HTWingNut's posting:
Sager NP9758-G / Clevo P750DM-G Early Initial Impressions
http://forum.notebookreview.com/thr...vo-p750dm-g-early-initial-impressions.782275/
"Keep in mind this is all just initial impressions out of the gate. After a good repaste and some time with the machine I will follow up with a detailed review and benchmarks."Last edited: Oct 3, 2015 -
^ yup, definitely L4 cache on broadwell. NO WAY a 4.1GHz broadwell chip beats a 4.2GHz skylake chip and even higher clocked Haswell chips any way else.
-
D2 Ultima, whatever the mechanism, the Broadwell CPU's have it, and the Haswell / Skylake CPU's don't, and the performance results show it.
If there is something that can be tweaked on the Skylake CPU to improve performance, it should turn up over time, it's still early days.
The other results of HTWingNut's are also outperformed, so perhaps it is a general tuning that will help.
If Windows 10 is to be blamed, purchasing Windows 8.1/7 can be a solution.Last edited: Oct 4, 2015 -
Broadwell has 128MB of eDRAM that functions as a L4 cache (other CPUs stop at L3 and then go to RAM). This is helping its Physics scores and SOME other CPU-related tasks.
There is not. Skylake will never outperform Broadwell in these instances, however it may for gaming and other situations where that eDRAM cache matters little.
I didn't look at what he did, but many CPU-based tasks might very well benefit from the eDRAM cache.
All firestrike scores used Windows 10, so that is not the cause of the discrepancy. -
D2 Ultima, all the other Firestrike scores were lower, and on Windows 10, the 5950HQ score was on Windows 8.1.
It would be interesting to see their results running on Windows 8.1. -
They wouldn't be as good as 7. 7 just flat out handles the CPU the best. I believe that 10 falls between 8.1 and 7.
The thing is, @D2 Ultima, I thought that the L4 was only on Iris Pro models? The i7-5700HQ is not Iris Pro.
TomJGX likes this. -
Win 10 and Win 8.1 have proven to give very similar physics scores in firestrike from what I remember.
But yes, OS parity is good.
In the case of firestrike, Win 7 is worse for CPU physics. 3DMark 11 however is best run on 7.
After doing more research, I am confused to all meowmix now. Broadwell seems to have some magic sauce, because not only are you right that the 5700HQ does not have its L4 cache, but all the Iris Pro Haswell chips like the 4980HQ tested DO have the same 128MB L4 cache for their Iris Pro chips, and Skylake's 6820HQ and 6920HQ both have the 128MB L4 eDRAM cache as well. -
I wonder if it is a Futuremark bug... I'd be interested to see how Broadwell handles the Catzilla physics test versus Haswell and Skylake.
-
HaloGod2012 Notebook Virtuoso
ill run catzilla shortly -
moviemarketing Milk Drinker
Is this only relevant for Firestrike Physics or do mobile Broadwell CPUs have an edge in other CPU-intensive tasks like video renders, encoding, image filters, etc? -
We don't need mobile to tell; desktop can tell too, since broadwell is the only product line whose desktop CPUs are designed exactly like their higher end mobile counterparts (eDRAM, Iris Pro GPU, etc)
-
HaloGod2012 Notebook Virtuoso
here is my 1080p catzilla bench: physics score is 1516
RESULT -
Haswell-H: i7-4750HQ and up have Iris Pro Graphics 5200 (40 EUs + 128MB eDRAM)
Broadwell-H: i7-5750HQ and up have Iris Pro Graphics 6200 (48 EUs + 128MB eDRAM)
Skylake-H: HD Graphics 530 (24 EUs, no eDRAM) -
Thats a 4770K though, not Broadwell.
-
You're right. Seems old skylake spec sheets lied; they said the higher end skylake mobile chips would have eDRAM and Iris Pro
-
moviemarketing Milk Drinker
The 5700HQ seems to perform a little better than 6820HK and 4980HQ in a number of synthetic benchmarks but still worse than 6700K and 4790K: Comparison i7-6700K, i7-4790K, i7-4770K, i7-3770K, i7-4980HQ, i7-5700HQ, i7-6820HKEthrem likes this. -
Broadwell kills it on SuperPi... Wonder what Intel did that made Broadwell so much better than Haswell and Skylake there
-
Because a 4790K and 6700K are at 4.2GHz versus a 5700HQ at 3.5GHz?
-
moviemarketing Milk Drinker
Not very familiar with this benchmark, but aren't higher scores worse? The score equals the number of seconds it takes to calculate PI up to a certain number of digits, no? -
Yep. wPrime is the same way.
-
HaloGod2012 Notebook Virtuoso
this was benched with my 5700hq, i dont own a 4770 -
*facepalm* I'm stupid sometimes. I don't use SuperPi lol. Now that I've made my dumb mistake for the day I'm going to go take a nap =P
When I went to see the result it said 4770K
-
HaloGod2012 Notebook Virtuoso
Maybe catzilla cant recognize the new cpu's? Was defintely benched on my GT72
-
Go look at the screenshot I added. The result is not the one you meant to share.
-
Man, my 4720HQ can at best do 10k on firestrike physics. That's a nice 20% boost for broadwell.
-
HaloGod2012 Notebook Virtuoso
let me re post it
sorry about that, that was a very old result on my account. Here is the GT72 result below:
CATZILLALast edited: Oct 4, 2015 -
Definitely seems like a Futuremark glitch.
http://www.catzilla.com/showresult?lp=454237|*|Result Details
Let me fire up my desktop since it doesn't throttle... -
My 4790K and 780 Ti. 4790K is at 3.7GHz (Catzilla won't read values below the stock 4GHz).
http://www.catzilla.com/showresult?lp=563635
It seems like Futuremark is favoring Broadwell for one reason or another.
3dmark06 - http://www.3dmark.com/3dm06/17813944HaloGod2012 likes this. -
HaloGod2012 Notebook Virtuoso
interesting. Either way, im happy this chip doesn't throttle...i was expecting the 5700HQ to sit at 2.7ghz the whole time like my previous HQ chips. -
Yeah I'm not slamming it any way, that's a pretty impressive chip you have there. I mean my 4940MX needs some tweaking to run at 3.7GHz without issue (default current limit is not high enough for it), it runs 3.4GHz under full sustained load (wPrime 1024) out of the box.TomJGX and HaloGod2012 like this.
-
I told you that was a problem with the laptops rather than the chips, and is why Alienware's current line can dive into an ocean for all we care.
-
HaloGod2012 Notebook Virtuoso
Yeah, the alienwares i tested had massive throttling.....wish i could go back to the days of the m9700 and m9750 models....good times
GTX 980 launched for notebooks
Discussion in 'Gaming (Software and Graphics Cards)' started by Cloudfire, Sep 22, 2015.