Actually, by far the most cost effective chip for overclocking is the E2160 for ~$70. It can overclock to 3GHz+ also, and has a 20% IPC advantage over the X2 5000+ BE.
Actually, by far the most cost effective chip for overclocking is the E2160 for ~$70. It can overclock to 3GHz+ also, and has a 20% IPC advantage over the X2 5000+ BE.
Yeah... gotta love those guesstimates.
Sometimes I think management will say anything to make us work harder.
lopri
Elite Member
- Jul 27, 2002
- 13,329
- 709
- 126
harpoon84: You really can't think those linked graphs are representative of CPU-bound games. Especially the Sim charts from that site are using Radeon X1900 for the test. I would take an FX60+8800GTX combo over X6800+X1900 combo any day. I believe others do the same if they're playing Flight Simulator X. (or any game for that matter) Again, we can make any game CPU-bound or GPU-bound. But what matters in the end outside of competitive arena (like professional gamers) is the GPU at any given time, budget, etc.
And the leaked presentation of Nehalem performance expectation says nothing about server performance. It's projected Spec_Int and Spec_Pf scores. Surely it's understandable that one can imagine that chart was created with server in mind, but it's still showing performance increase of Shanghai (from Barcelona) nonetheless. If Shanghai performs better than Barcelona in server environment, it's very possible that it'll also perform better in desktop environment in that they share similar architecture. (Harpertown performs better than Clovertown -> Yorkfield perfoms better than Kentsfield)
Of course it's just a speculation and I wouldn't believe it until I see it, but we were speculating in this matter anyway - and you've asked the base of the speculation so I provided. Again, it doesn't mean that I believe that chart from Intel, but that chart was quoted to project Nehalem's performance before (and made me very excited).
And the leaked presentation of Nehalem performance expectation says nothing about server performance. It's projected Spec_Int and Spec_Pf scores. Surely it's understandable that one can imagine that chart was created with server in mind, but it's still showing performance increase of Shanghai (from Barcelona) nonetheless. If Shanghai performs better than Barcelona in server environment, it's very possible that it'll also perform better in desktop environment in that they share similar architecture. (Harpertown performs better than Clovertown -> Yorkfield perfoms better than Kentsfield)
Of course it's just a speculation and I wouldn't believe it until I see it, but we were speculating in this matter anyway - and you've asked the base of the speculation so I provided. Again, it doesn't mean that I believe that chart from Intel, but that chart was quoted to project Nehalem's performance before (and made me very excited).
Actually, these graphs illustrate clearly games that can take advantage of faster CPUs. Who said anything about FX-60 + 8800GTX vs X6800 + X1900? You're the one thinking up such an irrelevant comparison. It goes without saying when CPU gaming performance is being compared the GPU has to be the same for an apples to apples comparison. The truth is that FS-X, amongst many other games, benefits from a faster processor, the evidence is there, if you choose to ignore it that is your prerogative.
All those CPUs are server based, and the context of the performance charts are obviously server (or, more accurately, SPEC performance) which relates more to server than desktop usage patterns.
What I am skeptical about are claims that the desktop variant, aka Deneb, will match or exceed Penryn.
lopri
Elite Member
- Jul 27, 2002
- 13,329
- 709
- 126
I clarified my post above to make my point clear. I am not denying the impact of faster CPU on gaming performance. As I said before, we can make any game look CPU-bound, not just sim games, if we want to. I have no intention to take this to age-old CPU vs GPU in games.
The current consensus is a 2.8 ghz core2 or 3.2 ghz amd x2 is enough to game happily with all titles this side of Crysis. From that standpoint the E2XXX core2 is cheaper, has more 'futureproofing' and is better accross the board. I'll admit that you can currently get by with an AMD CPU if you get a good OCs on it or if you spring for a $100+ chip.
With my own testing I saw HL2 frame rates go from 91 to 142 fps average moving from a 1.8 ghz to 3.33 ghz on a core2. The FPS drop was pretty linear starting at 3 ghz or so. That may not be perceptible but it is a 4 ms reduction in latency -- and with FPS every little bit helps.
So in conclusion, you can get by with an overclocked AMD CPU today -- but there are much better alternatives cheaper.
With my own testing I saw HL2 frame rates go from 91 to 142 fps average moving from a 1.8 ghz to 3.33 ghz on a core2. The FPS drop was pretty linear starting at 3 ghz or so. That may not be perceptible but it is a 4 ms reduction in latency -- and with FPS every little bit helps.
So in conclusion, you can get by with an overclocked AMD CPU today -- but there are much better alternatives cheaper.
I would say go with at least a 5000+ for performance sake in the chip, and for the motherboard I would go with the MSI K9A2 Platinum or DFI board...I have an MSI and DFI running AMD X2's and they both work great...it's all about the options, RAID, 1394, bla bla bla.
Problem is that an e2160 at 3ghz is slower than an e4300 at 3ghz because of the cache. The difference between the 1mb cache vs 2/4 ( e4x00 / e6x00 / x6xxx ) is pretty remarkable. The e2160 at 3ghz runs about as fast as a 2.5-6ghz 2/4mb version. A 3.2ghz ( oc or not ) AMD Athlon 64 X2 can pull 2.4-5ghz 2/4mb cache C2D speeds. 2.4ghz 2/4mb C2D performance is the sweet spot for gaming on a budget and ANY chip brand can hit that.
Phenom B3s and the .45nm on the horizon, there is no reason to really evangelize for Intel. If people want AMD its not the end of the world. My Opteron 180 4g doesn't feel that much slower than my e4300@3.0ghz 4g. One uses a 8800GS Superclocked and the other uses a 3850 512. I can play my games happily on both.
How is that a 'problem' exactly? If that is a 'problem' then the X2 5000+ BE has the same 'problem' since 65nm Brisbane X2s (like the BE 5000+) are also slower than the 90nm Windsor X2s (6000+/6400+) because of a smaller and slower L2 cache, as well as certain RAM dividers than underclocks RAM from the DDR2-800 spec.
Yes, we all know the E4x00 and E6x50 chips will outperform the E21x0 per clock, but at a significantly higher cost. An E6750 costs ~$190, an E4500 costs ~$125. An E2160 costs ~$70.
My point was that you can get significantly better price/performance (once overclocked) by getting the E2160 instead of the X2 5000+ BE. Despite the 1MB L2 cache, the E21x0 chips still have a ~20% IPC advantage over a Brisbane core X2, so when both chips are overclocked to 3GHz, the E21x0 is still ~20% faster overall.
An E2160 @ 3.4GHz beats a stock 2.93GHz X6800 in the vast majority of benchmarks. http://www.xbitlabs.com/articl...um-e2160_14.html#sect0
An X2 5000+ BE @ 3.3GHz barely beats a stock 3GHz X2 6000+ / 2.33GHz E6550. http://www.tomshardware.com/20...overclocker/page9.html
I like how you picked a random number like 2.4GHz and declared it a 'sweet spot' for gaming. I have an E4400 myself, comparing performance at 2.4GHz and 3.33GHz (my max 24/7 overclock) many games show improved performance at the higher clockspeed.
I'm not sure what B3 Phenoms and the 45nm shrink has to do with the discussion, but since we're talking about whats on the horizon, what about Nehalem? If anything the performance gap between AMD and Intel will grow over the next year, but that is a totally different topic altogether...
superstition
Platinum Member
- Feb 2, 2008
- 2,219
- 221
- 101
Give me one advantage of buying an AMD processor at this moment. Just one.
I was an AMD fan when Intel's crappy Netburst was around, but those days are gone. As far as I know, AMD doesn't even have an advantage in heat/power consumption anymore - the one thing it held onto until recently in specialized server workloads.
I was an AMD fan when Intel's crappy Netburst was around, but those days are gone. As far as I know, AMD doesn't even have an advantage in heat/power consumption anymore - the one thing it held onto until recently in specialized server workloads.
It is not "pretty remarkable", where you get that? Only in benches or apps that make use of the extra cache. Overall may average around 10% (encoding, multimedia may be 30-40%) while other apps or games near zero gain.
Intel has pretty much whomped AMD in multimedia. Current media transcoding advantage in many cases today are attributable to the current instruction set and SSE4 support. As more folks optimize for SSE128 this advantage will 'disappear'. AFAIK full AMD SSE4 support (and something called 'SSE5') will not be available until 45nm (or roughly late 2008 - 2009).
SSE5 is a whole 'nuther issue. AMD instruction sets are always at a competitive disadvantage because of Intel's market position and status as ""Chipzilla"". HOWEVER, if they demonstrate significant improvement (as has been shown in some cases) and early adopters gain a competitive advantage, Intel will be forced to follow suit.
OR, because of the preiminent market position of Intel, they could squash it like a little annoying bug ....
It's an interesting time for 'clock' wars, instruction sets and overall platforms. I hope the competition continues without undue manipulation. If Intel impedes the expansion of new instruction sets (to me) it would reinforce some of the claims that AMD has made concerning 'uncompetitive' market practices.
I'm sure others will see it differently
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 24K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter