Printable View
So as a gamer, how would buying a phenom benefit me?
Has there been any kind of tests carried out on the smoothness of a phenom vs a core 2? (Maybe somthing similar to the microstuttering tests carried out on sli/crossfire)
This isn't meant to flame, im actually interested. If it can be shown that phenoms provide a better game experience i would seriously consider getting one.
http://www.xtremesystems.org/Forums/...d.php?t=197423Quote:
Originally Posted by oohms
will shed light on what you seek. maybe
Oh christ, the 'smoothness' again..
I'm not going to say anything, haven't had or used an FSB based system since I replaced the XP with an A64, there might even be something to it but it sure sounds like make-believe to me. :rofl:
Using Blocksize 576kb to 2048 (to fit in the l3 Cache), Latency (Cycles) is around 43), Latency(ns) is around 17ns. For random accessing prefetcher should not work.Quote:
Originally Posted by savantu
Dont know if you know that L3 Cache has its own clock domain, so if you raise NB speeds you have an affect on the l3 cache speeds aswell.
Are you running the NB at 2.5GHz ? If so , that explains 17.2ns.Quote:
Originally Posted by Boschwanza
2750 MHz for the Cores, 2200 MHz NB DDR2 1100 Cl5, i cannot go higher.Quote:
Originally Posted by savantu
I made a little more testing on NB Speeds (Core Clock 2.6 GHz). From around 1625 to 1950 MHz you see absolutly no gains in terms of speed, with 2200 MHz it bumps up to almost 9% advantage (latency/cycles).
http://www.abload.de/img/nbuk01.jpg
I got an additional score from a friend fo mine with 2.6 GHz on the NB with following results:
Latency (Cycles)
Read Forward 16,05
Read Backward 16,28
Random 40,06
Pseudo Random 26,18
So all AMD has to do is to raise NB Speeds to a decent level.
I find it odd that the latency in cycles changes with frequency ; must be due to the asynchronous design of the L3.Quote:
Originally Posted by Boschwanza
Either way ; how did you get 17ns in random mode when the graph shows 44 cycles at 2.2GHz => 20ns ?
17 ns is for CoreClock 2750MHz and AOD max perfomance settings (red). The graph was made with 2.6 GHz and AOD performance mode (yellow). 44 Cycles is around 18 ns.Quote:
Originally Posted by savantu
I will re-run the test with performance mode red and 2750MHz, getting cycles and latency. Just give me a bit time.
Well, as I wait for my new motherboard Im running my Phenom for 2 months already at stock. Ive no problems what so ever. Although, Ive to admit that there's a clear difference between this and the Windsor F3 @ 3.5Ghz, that ran just a bit smoother due to higher clockspeeds, but that doesnt need Einstein to understand though:p:.Quote:
Originally Posted by oohms
So basically, if you need help to get interested for Phenom, just dont bother. It does its job great, later 9850BE's and all 9950BE's show a clear increase in OC yields compared to earlier Phenom's. So if you'd get a 9950BE with a SB750 board and get it minimal to 3Ghz you're settled already. Also you'll be ready for Deneb when it hits.
Although, I dont play everygame 24/7 though. But for example, Red Alert 3 Beta plays damn good. Only screen lag I get is from crap connections from other players. If the connection is very good it plays very well and yes, smooth. But then again what's the point of a 5Ghz CPU which doesnt run smooth:rolleyes:
Guys, I'm getting tired of having to delete inappropriate posts. If you can't get along with one another, then don't post. Otherwise you may be taking an unplanned vacation.
eh... whats inappropriate about saying that its not possible to notice a difference, because the mentioned engine has a framerate limiter and his "more smooth" perception is just imagination?
on enverything higher then a 8800gt C&C3 and all games based on the same engien output a flat 30fps.
BOOM Here comes sierra-bound BOOM ready or not BOOM how you like me now!
Nothing wrong, mine also get deleted. I'll change it a bit, but there was no offense in it...
The slower but smoother was funny, but now it's not. Seriously I can't believe this:
Then someone with a bit of intelligence will never buy a Phenom over an Athlon X2. Let alone C2D's and C2Q's.Quote:
Originally Posted by Rammsteiner
So you're recommending a system that WILL have less perfomance compared to similar priced AMD or Intel offerings?Quote:
Originally Posted by Rammsteiner
:rofl: here we go again with the slower but smoother. Stop trying to make the guy buy the Phenom giving him false reasons and data.Quote:
Originally Posted by Rammsteiner
If you want to game (well and if you want to do anything else with your desktop PC) you should only consider AMD if your budget is low. If not, forget about the slower but smoother thing spreaded by AMD fanatics, do yourself a favor and buy an Intel rig.
:rofl:Quote:
Originally Posted by Rammsteiner
But then again, what's the point of using the slowest space heater quad cores from AMD?
http://www.xbitlabs.com/articles/cpu...0_7.html#sect1
Seriously, what is this 'slow but smooth' FUD? Going by rammsteiners comments you'd think Intel CPUs microstutter in games or something. :rolleyes:
I had 3 of my AMDriod buddies try to show this to me on my HTPC with C2D E6600 and C2D E8400 that's not even overclocked at the moment. 2 of the 3 thought it was it runs so smooth. The HTPC is running Vista 32bit and 2GB of RAM lick-a-de-split (very fast) as well. They'd read this same BS on the AMDZone. If these systems stuttered we'd be here complaining and not making B2 Phenom like excuses.Quote:
Originally Posted by oohms
I've always thought the "it feels smother" argument was all in the head or a result of something that had nothing to do with the CPU. When Intel users were making the "smoother" claims during netburst I didn't buy it and I don't buy it for current AMD chips.
well, that was a bit different. intel cpus were smootherwhen there was a load on them, because of ht. not nearly as smooth as a real dual core, but it was still tangible. i do admit though, that my old opteron at times felt smoother than my q6600. Its not just me reading into this bull, under load, the q6600 feels more laggy than my opty did. It could be just something with the os, but it feels the same under xp and vista 32 as well. I'm not even attempting to make a guess as to why, just saying what I experience.Quote:
Originally Posted by MrMojoZ
What you are describing looks like an effect coming from much better memory subsystem Opteron has,at least when it comes to multitasking(IMC+low latency ram+dual core = very responsive system)Quote:
Originally Posted by cky2k6
I would hypothesize that it's because of the lower latency RAM accesses due to the IMC: lower latency memory accesses would always be "smoother" than high latency memory accesses with a band-aid prefetching solution, since the prefetching only works when the memory access patterns are predictable. However, prefetching technology should keep things almost comparable for most circumstances (there are always nasty corner cases).Quote:
Originally Posted by cky2k6
Cmon, are you serious?Quote:
Originally Posted by savantu
It's because the cores are at a fixed frequency while the uncore clock frequency is swept across different frequencies.
In games it could be possible for a processor to have a lower average framerate, but a higher minimum framerate. It can happen with various GPUs, and I don't see why it wouldn't be possible for CPUs also.
But with that said, I have never seen any reliable testing on the subject - even though it would be relatively easy. It always comes down to users feelings, which can be highly influenced by the individual's emotional investment and all sorts of other factors. Until I see concrete results that say otherwise, "slower but smoother" really sounds like grasping for straws. It would be easy to test, so why not see if it can be proven empirically?
If it really was true, why hasn't a single reviewer ever caught onto it? Are they all in Intel's pocket? :rofl: (I don't mean to be patronising, I halfway expect certain people here to suggest that so I just said it preemptively..)
JumpingJack spent a lot of time doing various different tests between Phenom and C2Q, trying to provide some actual facts to a 'discussion' (more like talking to a wall) with member 'gosh'. If there was a shred of empirical evidence to this 'smoothness' talk it would've been uncovered in at least one of those thorough tests.. The link is further up on the page.
indeed, avg frame for core 2 is most of the time (if not always) higher on C2 then Phenom, thought absolut min fps is a bit higher on phenom. But as said in the other topic, i take higher avg fps every time, over min fps which only represents one single point (1/10 or even 1/100 of a second). :yepp:Quote:
Originally Posted by bowman
And even when you only watch at min fps, its very game depended. Some games prefere C2 some Phenom.
Please dont try to discredit those who feel a smoother experience with one CPU over another telling them that its in there head because you dont know that. Its the same as the age old argument of how can you see or feel anything over 60 FPS.
I myself have built and used and benched all types of Intel, AMD, and VIA CPU's. I also like many others feel a smooth responsivness on a Phenom system unlike I have felt with others. I have no clue why or how but its there.
Edit: Id also like to add that I only notice this on 64bit OS. its not the same on 32bit.