Not with really highly threaded workloads...
Just registered to troll a little..?..
Not trying to troll at all. However I'm assuming an OC'd 3770K as I wouldn't see the reason to buy one if you're not going to OC it.
Not trying to troll at all. However I'm assuming an OC'd 3770K as I wouldn't see the reason to buy one if you're not going to OC it.
ShintaiDK
Lifer
- Apr 22, 2012
- 20,378
- 145
- 106
Highly threaded workloads aint enough. It needs to be highly threaded workloads with almost perfect scalability. And then we are down to something like <1% of all multithreaded applications.
You are right , this is an overclock dedicated i7 , so let s see how it fares
if overclocked to death at higher frequency than an overclocked FX....
You are right , this is an overclock dedicated i7 , so let s see how it fares
if overclocked to death at higher frequency than an overclocked FX....
OC'd to death? Lol. How about some gaming comparisons?
No , it could also be a few tasks that are each taking advantage
of only 3 or 4 cores.....
ShintaiDK
Lifer
- Apr 22, 2012
- 20,378
- 145
- 106
Add abit of serial code in the mix, and the performance of the FX drops significantly, nomatter the thread amount.
Changing the subject to escape the debate...??..
You clearly claimed the 3770K being better in an answer to a member
that did explicitly talk of multithread usage.
Excuse me for not knowing the parameters of said argument had been oh so narrowed. However, your reply showed the with both chips OC'd the 3770K still pulled ahead. Now if you have test results comparing them clock for clock when OC'd to shut me up, I will gladly take a seat.
Are we talking of multithread/multitask or of single thread/serial code.??.
Seems once something doesnt suit your spining theories you have
to bring again the single thread perf as the ultimate bench...
ShintaiDK
Lifer
- Apr 22, 2012
- 20,378
- 145
- 106
It would be much better if you could drop the bad attitude.
We are talking multithreaded applications.
A real life example where the FX8350 loses out due to abit of serial code in the multithreaded application is chess.
As you can see the application scales with cores, even 16 threads as shown here.
Dont let parallel and serial code confuse you with singlethreaded and multithreaded.
Amount of serial code also dictates the possible maximum speedup compared to cores.
Last edited:
You realize that at same frequencies the 8350 would be largely
as fast as the 3770K in the mentionned test ?...
You failed to see that in these MT tasks the 3770K has at least
300mhz frequency advantage but still , you re talking as if the frequencies
are reversed....
Talk of being blindly biaised...
in Australia from the equivalent of MC called MSY
Intel i3-3220 / i3-3225 / i3-3240 117 / 128 / 145
Intel i5-3330 / i5-3470 180 / 185
Intel i5-3550 / i5-3570 / i5-3570K 195 / 202 / 227
Intel i7-3770 / i7-3770K 295 / 332
AM3+ x8 FX-8120 /x8 FX-8150 /x8 FX-8320 /x8 FX-8350 165 /198 /187 /217
the K's doesn't support iommu which is a major PITA these days for a home server given Hyper V* and ESXi are so easy to setup and just so awesome and flexible so your forced to a 3770 which at that point you can quite easily pull back a fair bit of the single thread deficit as your 3770 is only getting another 200-400mhz from OC compared to 8320/50 which with sufficient cooling is rock solid on air at around 5ghz with minimal voltage increase. single thread cine @ 5ghz is around 1.33 which to me at that point we are well into "acceptable" single thread performance.
in my limited experience it appears the reviewers have been giving 8350 way to many volts and limiting there OC's because heat is really the only limiting factor keep it cool and it just keeps going.
The other big thing is if your doing real time trans-coding of 1080P H264 while trying to run your sidewinder firewall and IPTV servers as other guest hyperthreading just doesnt really cut it because transcoding gets good ALU utilization. So you end up having to give proportionally more resources to the other guests to keep them happy.
*HV requires win8 for non powershell management
. GG MSno your not, see how well i justified my argument..... juts like you did.
Lolz , only adobe ??...
The TR graphs i linked clearly show that the 3770K
although scaling well , doesnt scale/frequency as well
as the FX.
That said , we re talking of overclocked FX in this thread ,
so stay in the topic , please , and if people talk of Multithread
dont come responding with games or anything that is not
trhe subject , just spare us your marketing efforts...
Last edited:
Lolz , only adobe ??...
There's a reason I posted the link, I wasn't going to link 30 different URLs.
Amount of serial code also dictates the possible maximum speedup compared to cores.
The FX exceed your graph s 6 fold speed up for 8C.....
So what is wrong , FX or your graph..??....:sneaky:
It would be much better if you could drop the bad attitude.
We are talking multithreaded applications.
A real life example where the FX8350 loses out due to abit of serial code in the multithreaded application is chess.
As you can see the application scales with cores, even 16 threads as shown here.
Dont let parallel and serial code confuse you with singlethreaded and multithreaded.
Amount of serial code also dictates the possible maximum speedup compared to cores.
You know Shintai, Fritz is nowhere near the champion chess engines anymore...Houdini and Rybka are the current world champions of computer chess engines.
http://www.hardware.fr/medias/photos_news/00/39/IMG0039217.png
Torn Mind
Lifer
- Nov 25, 2012
- 11,643
- 2,654
- 136
The problem is ignorance from your end about Amdahl's law.The FX exceed your graph s 6 fold speed up for 8C.....
So what is wrong , FX or your graph..??....:sneaky:
http://home.wlu.edu/~whaleyt/classes/parallel/topics/amdahl.html
Surely , but anyway , congratulation for the wide range of benchs
in the four pictures you posted....
I know this estimation rule , it s 13 years old kid math level..
What i was adressing was the graph used by ShintaiDK that explicitly
show a 6 fold speed up for 8 cores while the FX exceed this value...
It comes down to, do you want something that's a LOT better at just about everything, or something that might be a tiny bit better at a very small number of things. Most people would choose the former, the other people like to simply support AMD and pretend that everything they do is optomized to run better on an FX processor.
83XX performs way better for what i what it to do then a 3770 and a K cant do it. Where do i fit in your over generalization. i buy the best $/perf processor for my needs, my desktops an I7, my servers an FX yet im just pretending
What do you use your server for mostly?
real-time 1080P transcoding for DLNA clients, IPTV, file server, Firewall ( sidewinder/MFE) , SSL VPN Termination (F5 APM), Cisco IOU/Dynamips/Olive Lab . the price difference between a 3770 ( K doesn't have IOMMU) and 8350 bought a Samsung 840 SSD that all my guests OS's run on.
Last edited:
TRENDING THREADS
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 10K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 7K
-
-
Facebook
Twitter