Quote:
Originally Posted by villa1n
Damn... I'm afraid that only ONE HD5870 could bottleneck x8 lanes in lynnfield :(
Printable View
Quote:
Originally Posted by villa1n
Damn... I'm afraid that only ONE HD5870 could bottleneck x8 lanes in lynnfield :(
What exactlly do you mean? Why would you use only 8 lanes with single card?Quote:
Originally Posted by golden_boy
Quote:
Originally Posted by kl0012
Forget it, bad sleep and no coffe :rofl:
I meant that i'm afraid of a single 5870 or 5870x2 could be bottlenecked in lynnfield platform (750) beacause i was thinking in that combo for my next build :D
regards
Quote:
Originally Posted by Clairvoyant129
Yeah silly Germans and their "language" :p:,a "small" and insignificant country with no knowledge of IT/computers,no IT market and poor journalists .
/sarcasm
BTW those two "silly" computer websites just happen to be among the best German computer related websites around with huge community behind them.The reviews used a large pool of applications,covering office ,multimedia and gaming and drew an average from those.You may like it or not,but they done a great job.
Yep, in his mind German websites are somehow less knowledgeable just because they use "foreign" language and their reviews are somehow "obscure"(despite the fact they used massive amount of apps in their reviews). :p:Quote:
Originally Posted by v0dka
Quote:
Originally Posted by gosh
You don't get it do you?
Why do you think those apps were created.. ? What comes first, the chicken or the egg? We as consumers use apps, play games or crunch, etc.. These apps are designed and made to benchmark all aspects of a CPU/GPU's performance.
People are showing you different graphs, to illustrate the point you are trying to make is rather minuscule. As there are other graphs testing the same micro-archetecture/parts that benefits having a L3 cache, etc..
You simply dismiss them, because you are not being honest with yourself.
DID I say they are less knowledgeable? I think I need to spell it out for you since you don't seem to be able to comprehend what I'm saying.Quote:
Originally Posted by informal
Since it is in a foreign language, I have no idea how they tested their platform. (yes I see your pretty little charts but I have no clue their testing methodology). Do you get it now?
No, I am a programmer and it is possible to check how the processor performs doing test apps. I am also developing applications that need performance.Quote:
Originally Posted by Xoulz
There is a "System setup" page in every review so those have it to.Be creative and click (for example) pages 3,4 or 5 and you'll see it don't worry. BTW,there is a thing called google translate and it does the job great for the quick'n'dirty translations if you really need to know the details of what and how they tested.Quote:
Originally Posted by Clairvoyant129
Yes, thats what i worry about as well, 5870 x 2 or 5870x2, which is what i wanted to build out...but with 8x pci-e 2.0 have enough bandwidth for maxed out 1920x1080 resolutions 16af 8aa etc .. Which is what i d like to play the new games at, and with eyevision, maybe grab another monitor as well.. so.. its pretty pertinent. :) I like the lynnfield because the oc results seem pretty easy to hit 4.0ghz 24/7 on air, with the right airflow.. versus the 920 where there's a bit of lucky chip factor involved still. That and the mobo's for x58 are about 70$ moreQuote:
Originally Posted by golden_boy
But even though you get x16 if you use a single slot, if you put a raid card or sound card or dvr card, whatever, on the second x16 slot it still knocks you down to x8 on that SINGLE gpu..i would not like being restricted like that.
You could just use a regular pci slot ^^ , i guess though a raid card is usually pci-e. Yeah for workstation type uses , not the ideal platform, but thats what x58 is for :PQuote:
Originally Posted by Computurd
Quote:
Originally Posted by Computurd
Or just use the pci-e slot wich dont split down to x8.. :rofl: there is boards that just use x16 / x4, or x8 / x8 and 8x / x8 / x4.Quote:
Originally Posted by villa1n
Most of the 1156 MB comming with x16 (x8x8)/ x1/ x1. Right now PCIe 2.0 x1 is enough for most periphery devices (including one SATA 6GB device).Quote:
Originally Posted by Migi06
These new Core i7 do appeal to me. The i7 860 is slightly faster than the old i7 920 and consumes less power which is another bonus and is priced closely. I may considering getting this with the P55 board. Does anybody know does the new Core i7 generates less heat than the older i7's?
yes the new i7's 860/870 do generate less heat and eat less also but they are not the ideal choice if you have the cash get a i7 920 instead.
Well what I can see the 1156 cpus reminds me more of the old stepping Bloomfields. They require more volt but also run a bit cooler. I run my rather crappy sample i7-860 at 4.0GHz 1.41v (bios) / 1.376v CPU-Z and I get 33~37C idle and 68~71C load with HT disabled and 73~75C with HT enabled in LinX with a CoolerMaster Hyper 212 Plus & Arctic Cooling MX-3 with stock fan @ ~1700rpm in 20~21C room temp. When gaming ~55C.
I could have afford X58 but I preferred choosing P55.
Q9550/9450 is quite old ... But I don't think you're serious :yepp:Quote:
Originally Posted by gosh
Intel Core i7-870 Review
http://www.nordichardware.com/Reviews/?skrivelse=559
No the guy is serious. There is actually a handful of these people over at amdzone who actually believe this. :shakes: Once in awhile they forget where they're at and start splewging out nonsense like he did and think that normal people will accept it like they do over at bizarrozone.Quote:
Originally Posted by Olivon
maybe you should read this document: http://software.intel.com/en-us/blog...t-prefetching/Quote:
Originally Posted by qurious63ss
Intel do have big focus on prefetchers and that makes the cpu fast in simpler loads, they don't work in heavier loads
Can you show some examples?Quote:
Originally Posted by gosh
do you not understand the document?Quote:
Originally Posted by gallag
Quote:
Originally Posted by gosh
No where does it say it can't handle heavy work loads.Quote:
it doesn’t help but likely won’t hurt performance either.
But just for fun, let me rub it in your face so you can have trouble sleeping at night.
Nehalem is better than Deneb, light or heavy workload. If Deneb is so much superior under 100% CPU usage, why is it that professionals and workstation users prefer i7?
Remember that it is a document from Intel. You just have to read about how prefetching works and how it affects the cpu and then it is simple to draw conclusions.Quote:
Originally Posted by Clairvoyant129
Workstations that do computational work should gain a lot from prefetchers. Even if there is very advanced algorithms that type of load is very simple to optimize and it needs very little code to execute.
Compare this when one advanced game need to calculate all information from a very complicated game scene. A lot of enemies (code for how they behave). AI for the enemies. Physics and code for that, explosions, wind, fire etc. All this needs very much code to be calculated.
If you just want to do a face which is looks very real with muscle movements etc. it is VERY complex.
Check 79 minutes in to this presentation from John Carmack when he describes what is complicated to do in games.
QuakeCon 09: John Carmack Keynote
Just calculating triangles is a very simple task, so simple that now with DX11 the gpu can do it (tesselation). In simple scenarios in games where streamlined computation doing triangle caluculations, prefetchers on i5/i7 would work very well for better performance. I also think that is the reason why i7 in some performance test is so fast and in others it is loosing, the L1 and L2 cache isn't that big, it is designed to be fast and work well with prefetchers but when it doesn't work it isn't that fast anymore. There AMD is faster.