-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
BF4 Retail CPU scaling measured
- Thread starter CAT-THE-FIFTH
- Start date
Very True and you have a free heater too, which in this cold weather is a very big advantage AMD has over Intel.
Associate
- Joined
- 31 Mar 2012
- Posts
- 1,737
It does. It reduces bottlenecking quite a bit and improves minimums. In my case with the 7990, the 8320 bump from 4.6 to 4.8 GHz improved minimums by 10%.
Milleage varies of course but that can be the difference between one's perception of "doesn't run smoothly" to "runs nicely"
Edit: see top of page 10 for my results.
I had an i5 2500k at 4.3ghz during the beta and gpu usage on my 670's was on average between 65-75% and the framerate when looking over all the shanghai map on a full 64 player server was ~60. Minimums went down to low 50's. My i5 was maxed out on all 4 cores 100%. I then upgraded to a 3770k at 4.5ghz after the beta and now GPU usage averages more like 80-90%, minimums in Shanghai are low 60's and looking over the whole map on a 64 player server has the framerate at ~75.
Pretty large improvement ( I would say 25-30% increase in framerates overall)
Crysis 3 is the biggest improvement though. In that Welcome to the jungle level with all the grass my minimum with the i5 was 28! With the i7 in the same seen it doesn't go below ~60! Massive improvement and I am hoping that is a sign of things to come once the next gen consoles are out and more and more devs code for 8 cores/threads.
Pretty large improvement ( I would say 25-30% increase in framerates overall)
Crysis 3 is the biggest improvement though. In that Welcome to the jungle level with all the grass my minimum with the i5 was 28! With the i7 in the same seen it doesn't go below ~60! Massive improvement and I am hoping that is a sign of things to come once the next gen consoles are out and more and more devs code for 8 cores/threads.
Last edited:
swear you do it on purpose .
its not a sign of tihngs to come you are using two gpus with effect the results massively sometimes.
single gpu i5/i7 no difference yet again.
watch first year consoles are out how much it changes. it wont !
its not a sign of tihngs to come you are using two gpus with effect the results massively sometimes.
single gpu i5/i7 no difference yet again.
watch first year consoles are out how much it changes. it wont !
Nothing to do with single or dual gpu, just with how many frames your graphics solution can put out. An i5 doesn't cause a bottleneck with most single card solutions because most single card solutions can't output more frames than the minimum that the CPU can. However, a 780TI is already snapping at the heels of my 670 sli in performance terms so that single 780TI will most likely be bottlenecked by an i5 in some games/scenarios.
I am not quite sure you understand fully how a bottleneck works.
Also, posting benchmarks for BF4 on an EMPTY server is totally useless. That is probably one of the least demanding CPU scenarios that you can look at! An empty map with nothing going on, no AI, no other players moving about affecting the environment e.t.c
I have seen it with my own eyes. Crysis 3 is the most striking difference though, as I said. If you have a powerful graphics set up, you will need an i7 to get the best out of it in some games and I suspect more and more games will start to use 8 cores/hyperthreading more effectively over the coming years.
Yes, if you have a single 670,7950,680,7970 the graphics card probably wont be able to output more frames than an i5 can put out but once you go sli/xfire with those cards or step it up to a 290x or 780Ti your graphics solution will be able to put out more frames than an i5 can manage ( again talking about these few specific games)
Last edited:
The thing is DG with ipc improvements small to non existant from one gen to the next and the Mhz/Ghz war all but over due to neglect, primarily because of power consumption and the percieved need for efficiency. What is there left to enhance the processing power of modern era cpu's.
In short we need more physical cores and if they coded effectively more virtual cores.
When gpu's become more powerful our 4 core Ivy's and Sandies will not cope and infact right now when in X fire and I use ultra and run a second monitor for battlescreen my cpu is at 100%. Yes its crossfire and yes I'm using a second monitor in ultra, but we need to be honest that if I had an i7 it wouldn't be 100% cpu utilisation it would be 75-80%
Games wont advance with 4C or even 4C 4T only, we need more cores (and code to utilise those cores) and likewise if threading works correctly more threads.
BF4 already uses 6 cores or with HT on 12 threads on a Sandy bridge e so we know that skillful Conscientious developers can utiltise extra threaded hardware.
Intel has held us back for the past 4 years in core count and general technology due to lack of competition from AMD. If AMD had produced a stonkingly fast 8 core, so too would intel.
Gaming developers would push performance to another level and games would be much better.
So in short for extreme duty gaming and gaming in ultra with all the bells and whistles you need multi gpu's and at least an i7 or Hex core intel.
Remember I have the Ivy I5 and hate giving Intel my money. So I say we need an i7 with a heavy heart....
Last edited:
jono i have benched i5s against i7s upto 4.6 on bf4 in mp i play with the guys every night in single cards and sli and crossfire
i5 will be fine with any current card even a 780 ti it wont bottleneck it at all.
the reson you notice better preformance is you have two cards that is all.
also there are numerous benchmarks now not one on a empty server.
i5 will be fine with any current card even a 780 ti it wont bottleneck it at all.
the reson you notice better preformance is you have two cards that is all.
also there are numerous benchmarks now not one on a empty server.
Having two, three, four cards doesn't matter, it is the overall performance(framerates) that your graphics solution can output!
Links for 64 player server benchmarks?
Soldato
- Joined
- 21 Jul 2005
- Posts
- 20,710
- Location
- Officially least sunny location -Ronskistats
I am not quite sure you understand fully how a bottleneck works.
Also, posting benchmarks for BF4 on an EMPTY server is totally useless. That is probably one of the least demanding CPU scenarios that you can look at! An empty map with nothing going on, no AI, no other players moving about affecting the environment e.t.c
The guy you quote Jono posts a lot and I struggle to take posts seriously as the grammar leaves a lot to be desired.
Looking at BF4 (as the thread is what it is about) then clearly the application to multi-cores has yet to reach it's full potential. Banging on about single cards and intels with only a fud benchmark that is flawed does nothing to the discussion really.
Soldato
- Joined
- 21 Jul 2005
- Posts
- 20,710
- Location
- Officially least sunny location -Ronskistats
This irritates me. ALXAndy would regularly get gunned down when he used benchmarks in SP from others on here. Take no notice.
Well can you link them again? I mean actual populated servers. An empty map/server is an utterly pointless way to benchmark CPU's in battlefield 4 multiplayer.
The guy you quote Jono posts a lot and I struggle to take posts seriously as the grammar leaves a lot to be desired.
Looking at BF4 (as the thread is what it is about) then clearly the application to multi-cores has yet to reach it's full potential. Banging on about single cards and intels with only a fud benchmark that is flawed does nothing to the discussion really.
Indeed.
Note:
Battlefield 4 utilises all the threads on intel cpu's. Infact up to 12 threads simultaniously on a hexcore. Whether they use all threads equally and efficiently is another story. Thread workloads obviously vary depending on whats being processed or allocated to each particular thread by developer code for example. I've no doubt Dice have got a massive degree of optimisation work left. Who knows there might even be a massive amount of junk code that needs removing fromt the game that will free up cpu resource. (This would help i5's out massively)
Other than that, the game runs like a bag of excreta sometimes and still crashes randomly. So they need to stop the crashes before they sort the cpu optimisation out.
Last edited:
trying to get official figures from frostibte engine people. then we can stop this debate.
also players i play with everyday on mp bf4 have i5s and i7 and some have the exact same gpu cards in single configuration and crossfire
. as we are playing the exact same maps at the exact same time side by side you can guage whether there is any differences.maybe ask someone like 8pack to drop a i5 in and test a 780ti and a 4770 see the difference
trying to get official figures from frostibte engine people. then we can stop this debate.
also players i play with everyday on mp bf4 have i5s and i7 and some have the exact same gpu cards in single configuration and crossfire
maybe ask someone like 8pack to drop a i5 in and test a 780ti and a 4770 see the difference
A 780ti wont bottleneck an i5 ivy/haswell it isnt powerful enough.
From what I can tell it takes circa 140 fps in ultra 4xmsaa at 1920x1200 to virtually bottleneck the i5's. (95%-99% CPU UTILISATION)
I know that running a battlemonitor cost me 10-15 frames and gives 100% cpu usage and adds instability to my system. (103oC at 4.61Ghz)
Fact is, my i5 2500k could only manage about ~60fps in intensive scenes/looking at a lot of the map in Shanghai. My 3770k does more like 75-80. The IPC and 0.2ghz bump over my 2500k would only account for a 10-15% increase but i'm getting more like a 25-30% increase so the hyperthreading is definitely doing something right. Even with the 3770k my GPU's still aren't being used at 100% on a full 64 player multiplayer map.
Last edited:
If your i7 cpu isnt 100% (which it wont be 75-80%) then your gpu's should be at 100% utilisation each unless you use vsyn. Something isn't right Jono.
Just the way it is with Battlefield. I don't think I have seen anyone who gets 100% on both cards in an sli/xfire configuration on a full 64 player map in BF4 ( or 3 for that matter).
Single player both GPU's run at 95%+ utilisation.