System Builder Marathon: Overclocking Day 3
Forum Overclocking : General Discussions - System Builder Marathon: Overclocking Day 3
http://www.tomshardware.com/2007/0 [...] index.html
We look at what components make an ideal budget overclocker's system. With a few bucks and some guts, you too can have a low-cost rig that runs with the big boys.
Wow, you couldn't have used a worse overpriced cooler. BTF90 easily dominates that cooler and costs only $40 w/o rebate less with. Should have gone with a fully enabled Conroe instead of the gimped E4XXX. I would have liked to have seen results with a E6600. And maybe to save money get a HD2900 or 8800GTS640Mb.
Ok, I understand that the C2D is the fastest CPU, however the best value? I've been over clocking since my 486 DX2 50! I read toms back when the Celeron 300A was hot... mine was not so hot. I read toms back when the Celeron 1.6A was hot.. mine was good. I believe overclocking is a great way to get a little more out of your money.
Here's my issue, toms has always gotten great results. To get the results posted in this article you need to pull great memory, a solid motherboard, and a solid CPU. If any one of those three comes up short, your hosed. Ok so you buy better memory, or MB, or CPU, there goes the budget.
Which brings me to my point. In all this C2D talk, people seem to have forgotten the X2 6000+ for 160 bucks! In test the cpu was 120 bucks and the cooler 55! That's 175. So with zero effort the out of the box X2 steps in with solid performance. Tweak, even 10 - 20% which is MUCH more likely to happen and you just beat this heavily overclocked system.
Here's a little something for you all to chew on... My Celeron 1.6A, which was OC'd to 2.4G, served me well for about 3 years. The system was used in an office for about 8-10 hours a day M-F. Well, BSDs became common place. Took me a week of testing to identify the CPU failed! I replaced PSU, Memory, Video, and motherboard before I suspected the CPU. Have you ever tried to find a replacement part for an old system? Talk about tough! Moreover, talk about expensive! If you consider time and effert used/lost in the trouble shooting, I just paid for an initial high end system!
I personally like to retire my OC'd systems to lesser use. It is very bad when a retired computer fails. So, this "75% overclock" way risky. If you get the right stuff to hit the 75% mark, how long will it last. Will your budget be cut short because of the catastrophic failure and now your wife AND you need a new pc?
I am greatly saddened that all these "high end" websites consistently overlook the the AMD value. A 6000+ with a very safe 10-20% OC is a budget system with incredible performance and VERY low risk.
So how about Tom's, how about some review of a budget AMD system?
-Peter.
Message edited by PeterHighlander on 07-19-2007 at 04:46:39 PM
| chewbenator wrote : Wow, you couldn't have used a worse overpriced cooler. BTF90 easily dominates that cooler and costs only $40 w/o rebate less with. |
Wow, you're the king of drama, considering there's a measly 15$ difference between the two. Also, please produce some evidence that the BTF90 'easily dominates' the XP-90... and no, three degrees doesn't count as 'domination'.
| chewbenator wrote : Should have gone with a fully enabled Conroe instead of the gimped E4XXX. I would have liked to have seen results with a E6600. And maybe to save money get a HD2900 or 8800GTS640Mb. |
Both of those points are covered in the review, your criticism might be more meaningful if you read it first...
Message edited by Cleeve on 07-19-2007 at 04:49:41 PM
| PeterHighlander wrote : Ok, I understand that the C2D is the fastest CPU, however the best value?
|
Hi Peter,
I certainly don't have anything against AMD... hell, I was the one of the folks who insisted that the previous budget marathon build used an Athlon X2.
But I don't think an OCd 6000+ would have beat an OC'd Core 2 Duo like our e4300, even with high-end cooling.
The Core 2's are just really good overclockers.
plus, $160 get's alot closer to the e6600's price, and the 6000+ wouldn't hold up well at all against that CPU in an OC contest, methinks...
Message edited by Cleeve on 07-19-2007 at 04:53:31 PM
Clock for clock the 6000X2 doesn't match up with the Conroe, and their ability to easily achieve 50% overclocks. And regarding the XP-90, this thing is old, so old that they even mentioned the fact in the review. As such I couldn't find a modern review that contained the BTF90 that also contained the XP-90. Fact of the matter is, there are so many better coolers out there. They should have used something at least that the average consumer can get a hold of and outperforms the Xp90 to achieve a good air overclock.
Message edited by chewbenator on 07-19-2007 at 05:02:28 PM
Something smells a little fishy to me on Page 5 - I thought only the "extreme" versions of the core 2 lineup had unlocked multipliers. So any ideas how they changed it in these tests?
The only way that comes to mind is if Tom's used an Engineering Sample of the e4300 that includes the unlocked multiplier (which isn't available on retail/oem chips)
Good article cleeve. Like the details of your trials.
Unlike many here, I appreciate stuff like this. Not b/c it tells me what I can do (we all already know the core2 oc's nicely) but b/c it is a FUN article. Who cares if it is an "old" cooler or engineering sample? They still work don't they? (I also agree, 3 degrees does not warrant buying a new cooler. If I had one on hand that was only that far behind, I would use it too) Honestly, you did well mentioning the age of cooler and other components but apparently too many peeps are just not satisfied w/ that.
I enjoy stuff like this, it is fun for me to read and a nice break at work. Keep it up man.
rock on.
Message edited by sojrner on 07-19-2007 at 05:33:45 PM
Reply to sojrner
| menetlaus wrote : Something smells a little fishy to me on Page 5 - I thought only the "extreme" versions of the core 2 lineup had unlocked multipliers. So any ideas how they changed it in these tests?
|
The asus P5B motherboard allows you to use lower multipliers with a retail/oem Core 2 Duo CPU.
I purchased both CPU and mobo from a local retailer...
Message edited by Cleeve on 07-19-2007 at 06:04:39 PM
| chewbenator wrote : As such I couldn't find a modern review that contained the BTF90 that also contained the XP-90. |
So you're saying you just guess it would 'dominate' the XP-90 then?
| chewbenator wrote : They should have used something at least that the average consumer can get a hold of and outperforms the Xp90 to achieve a good air overclock. |
Once again, this was mentioned in the review... and the Thermalright SI-128 was mentioned as an alternative.
I see your point, but I think you're overreacting a bit on the cooling front for a $15 dollar difference, while making wild performance claims you have no data to support.
| Cleeve wrote : The asus P5B motherboard allows you to use lower multipliers with a retail/oem Core 2 Duo CPU.
|
Ahhh... somehow locked doesn't seem like the right term for the multiplier then.
Saying the multiplier is limited to 9x (for e4300) would make more sense IMHO... but oh well
It was a good article, but since when is $1200 a "budget" system, lol. To save some dollars, it would have been nice to see you use a cheaper cooler like the ACFreezer7 ($22 shipped from ewiz.com). I like the CPU choice, and I bet with a different motherboard, a higher overclock would have been reached. I do, however, understand that you had to "use what you had" at the moment. In my eyes, a [b]budget[b] overclocking system would be much cheaper (3600+, ACFreezer, 7900GS/x1950Pro) but would obviously not compete with the high-end rigs... but what true budget system will?
Nonetheless, kudos on the article.
Home: E4600@3.6Ghz|AC7|P35|4GB|640GB|8800GT|Vista64|24" P-MVA
Work: Phenom9500@2.5Ghz|AC64|690G|3GB|500GB|8600GT|XP|2x22" TN
Reply to TSIMonster
| menetlaus wrote : Ahhh... somehow locked doesn't seem like the right term for the multiplier then.
|
Yep, its mostly up to the motherboard manf.
I can lower the multiplier on my ASRock board as well. This comes in handy on the oddly multiplied brisbanes (9.5x on my 3600+)
Home: E4600@3.6Ghz|AC7|P35|4GB|640GB|8800GT|Vista64|24" P-MVA
Work: Phenom9500@2.5Ghz|AC64|690G|3GB|500GB|8600GT|XP|2x22" TN
Reply to TSIMonster
so $22 vs $55 would make that $1200 better? lol
Whole thing seemed pretty clear to me. Cleeve explains each choice in the article. Dunno why so many are having issue w/ it. Just read it for what it is man.
meh.
Reply to sojrner
I don't know what kind of work you do, but to me $33 is a lot of money man. I'm not dissing the article, I like it. Just my opinion of budget is different, I guess.
Home: E4600@3.6Ghz|AC7|P35|4GB|640GB|8800GT|Vista64|24" P-MVA
Work: Phenom9500@2.5Ghz|AC64|690G|3GB|500GB|8600GT|XP|2x22" TN
Reply to TSIMonster
well, he refered to the previous marathon where they defined a "budget" system as under $1k. Using THAT definition (regardless of what you or I think of as a budget system) he then says the 8800gtx put him over that but he needed it to compare w/ the big dogs. (which was the goal of this marathon)
IMO it was all explained really well.
As for the cooler... If I had a $55 cooler on my shelf or could BUY a $22 one that may or may not gain me some cooling... I would stick w/ the one on my shelf. (which is what they did if you read between the lines a bit
)
Reply to sojrner
| Quote : But why didn't we choose the e6600 processor instead? For $100 more the e6600 has a sweet 13x multiplier and double the cache for extra performance. In addition, the e6600 has been known to reach speeds of more than 3.5 GHz. |
Uh... my C2D E6600 has a 9x multiplier... what E6600 are you using???
IMO, the cooler wasn't even the limiting factor. We have e4300s down here clocking 3.15G on the stock cooler. If the article didn't mention that they tried to jump the strap by going straight to 400+ fsb, I would think the issue was with the NB strap.
I'm currently fooling around with a 945GC-A2 chipset board and what I noticed is that I can't get beyond 350fsb on the e4300 when using a Sata HDD (amazingly enough, I can do 370fsb on a B2 stepping e6300). I plunked in my old trusty IDE drive and broke 370fsb on the e4300 immediately following the replacement. I will need an after-market cooler before I can shoot for that 100+% OC.
The situation above tells me several things: Sata consumes more power than IDE - that may be a factor; Sata controller is on the NB, IDE on the SB (?) on the 945G-AC which may or may not be a factor.
Message edited by jestersage on 07-19-2007 at 10:51:53 PM
Good article Cleeve, as usual. Too bad you couldn't include the high-res scores, hopefully they will be there in the final comparison. I'm sure that this system will get soundly beaten by the higher-end systems (at least in the graphics benches and especially in the higher resolutions), pretty much solely because of the 2 vs 1 video cards. That being said, it's nice to see that the CPU doesn't really make much of a difference (even less if you went with an e6600) and that even with a cheap e4300 things become GPU bottlenecked at average (to high) resolutions (1600x1200).
| Eurasianman wrote : Uh... my C2D E6600 has a 9x multiplier... what E6600 are you using??? |
Wondering the same??
Eurasianman wrote :
|
Ouch! Serious brainfart on my part there. I'll get that fixed...
I not a gamer but got into overclocking because I do a lot of home video editing. When I am making a video, I run my e4300 at a stable 3.51 Ghz (FSB 390Mhz) on my Gigabyte GA- 965P-DS3 v3.3 m/b. I couldn't get past 388Mhz FSB until I added a Thermaltake CL0034 NB cooler. The cost of my upgrades (m/b, cpu, NB cooler) was about $270 which seems reasonable for me given the performance increase.
I personally have the XP-120 which is basically the successor to the XP-90, see the similarity of the names? I also personally own the BTF90. From my own hands on experience the BTF90 outperformed the XP120 by a good 8C. Considering that the XP-120 is a larger cooler than the XP-90 and also newer to the market, being its successor I would expect it to outperform the XP-90. So, from my own experience the BTF90 trumps the XP-120 and should therefore trump the XP-90.
I had to resort to the XP-120 because it was a decent cooler that fit in my MountainMods Bob Slay case. These were tested against each other with three remounts per cooler, each producing within a degree difference of the other. So, saying I have no data to support this is bologna I just didn't have time to produce this post this morning. I can produce TAT and Core Temp screenshots if you would like, but I would have to remount the BTF90 because I have lost the previous test's screenshots due to a system re-format.
The BTF90 also has the advantage of a Bolt through mounting system so that it receives proper pressure and contact. From my own personal tests I have found it to outperform a newer cooler from the same company and product line. Newer does not always mean better, but due to the increased size of the XP-120 and therefore increased amount of fins and heatpipes I do believe I have made the right judgment.
Nice OC on your e4300!
| chewbenator wrote : Newer does not always mean better, but due to the increased size of the XP-120 and therefore increased amount of fins and heatpipes I do believe I have made the right judgment. |
That's great for you and I'm glad you're happy with the BTF90.
However, I still think the sensationalism of your initial response far outweighs the importance of the $15 difference in the review.
But whatever floats your boat, dude.
This is a good overclock, still I've had the 4300 stable at 3.4 GHZ with the gigabyte 3d cooler: GIGABYTE GH-PCU22-VG 105mm 2 Ball bearing.
but I used a EVGA 122-CK-NF66-T1 LGA 775 NVIDIA nForce 650i Ultra ATX with a good power supply
the cooler does an extremely good job of cooling the northbridge as well. this is all I was worried about, given the core 2's are not hot heads.
I am typing to you on a 1.86 GHZ core 2 6320 with 4mb cache running at 3.56 GHZ it can do a 1m pie crunch in exactly 14.5 seconds. (on the gigabyte cooler) Of course it makes a large difference in price to run a 8800gts 320mb and seems to score the same as the 640 mb version but saving 200 bux... seems worth it as I've not found a game that runs choppy with this setup. I used better ram... seemed to make the difference but with how ram prices are going down it's somewhat affordable to get some XMS2 1066 mhz ram memory timings didn't make a noticable difference.
the 650i ultra is a board that deserves some light shed, the stability of it is better than anything I've seen I've had the FSB set to 2033 MHZ and it booted windows with the 6320 in it. the motherboard only costs 100 dollars.
here is a screenshot
forums.extremeoverclocking.com/attachment.php?attachmentid=105538&am p;am p;am p;am p;am p;am p;d=1184015492
well let me know i can email you the parts to it. It seems the cheap Gskill Ram pc6400 (800MHZ) ram overclocks almost the same as the xms2 it's pretty darn stable
950 bux is the complete cost w/out xp
Message edited by Hargak on 07-19-2007 at 10:44:08 PM
every individual proc is different. Regardless of cooler, if the same one is used on two procs from the same bin, on the same mobo you can still get different ceilings on each one. Just the nature of a cpu man.
but you are right, it is a good oc... and so is yours. 8)
Reply to sojrner
not always true, I work for a mid sized custom computer builder and we see all core 2's able to hit at least 3.2 90% of them hit 3.4 motherboard power supply and ram are the main factors, yes liquid nitrgen will help alot but we're talking about budget. 27 dollar cooler, that's quiet and cools the northbridge
the validity of that statement not withstanding, you said 90%... that means not all of them hit 3.4.
Cleeve even states in the article that he wanted 3.4-ish... perhaps he got one of your 10% slugs?
regardless... not wanting an argument over theoretical statements for all core2's. Rock on.
Reply to sojrner
i did say 90%, no I'm not a kiddie flamer, I'm a professional builder, honestly I haven't seen any of the 10% I think the 10% are the ones I've heard of that didn't because they were using a 965 chipset or lower as opposed to 6 series or better nvidia, EVGA though... great cheap good product looking through some records we were able to get a 6700 core 2 to 5 GHZ wasn't me though
i meant 3.4 on my original post. which was done many times, only on the 650i ultra and 680i
Message edited by Hargak on 07-19-2007 at 10:43:10 PM
| Quote : from hargak: the 650i ultra is a board that deserves some light shed, the stability of it is better than anything I've seen I've had the FSB set to 2033 MHZ and it booted windows with the 6320 in it. the motherboard only costs 100 dollars. |
I like! Nice OC, too!
Anyway, back to topic:
1. I think the P5B, although representative of the 'budget OC board', is still at the top end of that list.
2. I am not convinced the e4300 is the limiting factor to the OC
budget oc board (in my opinion also the best for non-sli systems) evga 650i ultra.
again 96 dollars... opposed to 115 for a "mediocre"? board no offense on 965 just doesn't float my boat, we sell them here and it's very common to not hit 3.4 with them I believe that same cpu in the EVGA will hit 3.4 and be stable, the OC options in it are amazing.
I'd say an E6320 myself.
Were talking budget in this column, hense I would pick the 8800GTS for $279 as opposed to a $500 8800GTX. Maybe thats just me?
I'm sorry, but in my experience that Wintec stuff is trash.
I was pleasantly surprised by G.skill, Super Talent and Geil DDR2-800 modules. Micron D9 > most
The raidmax case?--sketchy; I'd take a Soprano or other decent mid-tower
I"m not sure about the P5B; by glance @ newegg I find an IB9 for $87. The one I have handles 400mhz FSB and it has a MIR for $30.
On the other hand, the P5B-E I had did 425mhz FSB stable...so is the possible $50 more worth the extra mhz to you???
$1100 is a pretty solid rig. Once we start mentioning sub $1000 or less, then I"ll consider that "budget"
still a fine article though : <3 you cleeve
Message edited by raven_87 on 07-19-2007 at 10:50:42 PM
| Hargak wrote : This is a good overclock, still I've had the 4300 stable at 3.4 GHZ with the gigabyte 3d cooler: GIGABYTE GH-PCU22-VG 105mm 2 Ball bearing.
|
Raven you echoed me, *high five* I'm with you as well
I find the 50 dollar cooler master cases with 3 120 mm fans do nicely not to mention the one with the plexi window also 49 dollars and still a sharp gamer looking case
Message edited by Hargak on 07-19-2007 at 10:55:38 PM
I do believe I'm flooding, but I had another quick tip, if yuo notice in my screenshot it only has 512 mb ram. there is a reason, there is only one chip in there, I have since moved to 1GB x2 but in this case, bottlenecks can cause bottlenecks, ie, if you think your ram could be the reason stability dropped down and you have 2 ram sticks, try pulling one, running it testing, and then swapping you can sometimes isolate which is faster and get the absolute overclock and still be stable.
The E6320 I had maxed out on a DS3 @ 450mhz FSB...
465x7 = 3255mhz on air. I was using 2x1GB sticks of Geil DDR2 - 800 @ 5-5-5-15-25 2.1v
The $$ you save on choosing the 8800GTS lets you exercise the option of the 4mb chip. So really that would be my ideal" budget setup.
what's your voltage at?
Sojrner, that's something i always wonder about when reading overclocking articles...exactly how lucky was the site with the cpu they bought/received? I'd love to see an article where multiple same model cpu's are overclocked, preferably all sourced from different different seller's to make it really random. The more that could be tested the better but lets say 10 would give a fair idea, for example test the max overclock of 10 e6600's using the same rig and see what the min/max results were.
Sometimes you see replies to posts where people say 'oh you should be hitting 5GHz with your dog doing warm farts on your rig with that cpu' and you can't help but think bollox
Carod
^^ I agree. I would enjoy an article like that, Just make sure the motherboard is up to the task. (As in, buy the know "best" overclocking board out there)
Home: E4600@3.6Ghz|AC7|P35|4GB|640GB|8800GT|Vista64|24" P-MVA
Work: Phenom9500@2.5Ghz|AC64|690G|3GB|500GB|8600GT|XP|2x22" TN
Reply to TSIMonster
| carod wrote : Sojrner, that's something i always wonder about when reading overclocking articles...exactly how lucky was the site with the cpu they bought/received? I'd love to see an article where multiple same model cpu's are overclocked, preferably all sourced from different different seller's to make it really random. The more that could be tested the better but lets say 10 would give a fair idea, for example test the max overclock of 10 e6600's using the same rig and see what the min/max results were. Carod |
actually, my dog assists in "water-cooling" my system... I modded the case to look like a fire hydrant to keep him motivated. I can beat out NO2 cooling that way.
but I agree... that article would be cool. Dunno how practical or feasible for a site like tom's to do it though in terms of how many "new readers" it would draw.
come to think of it... might actually be worth it in that way. Maybe if more of us suggest it they would warm up to it...
...as warm as a dog fart probably.
Message edited by sojrner on 07-20-2007 at 07:11:01 AM
Reply to sojrner
| chewbenator wrote : I personally have the XP-120 which is basically the successor to the XP-90, see the similarity of the names? I also personally own the BTF90. From my own hands on experience the BTF90 outperformed the XP120 by a good 8C. Considering that the XP-120 is a larger cooler than the XP-90 and also newer to the market, being its successor I would expect it to outperform the XP-90. So, from my own experience the BTF90 trumps the XP-120 and should therefore trump the XP-90.
|
Well woopdedo! Guess I screwed up when I paid $AU40 for my gigabyte neon 8 pro, it only cools my 3700 to sub 45C temps @1600rpm. In fact, it only managed 18C @3000rpm, what kind of cheap crap is this? Better go and pay for a new cooler
[/sarcasm]
Message edited by randomizer on 07-20-2007 at 08:12:52 AM
I haven't been OCing Intel in a while. I know on my X2 system(s) I can OC the CPU and reduce the memory multiplier. This hits performance a bit, but the onboard memory controller and less expensive ram is a reasonable trade off in my eyes.
Now what about C2D? Can you increase the FSB and decrease the memory multiplier? If you pulled a "bad" lot of DDR2800 that only did the rated speed, what does that do to your performance?
Seriously, I've been down this road many times, banking on a "good" overclock will always leave someone coming up short (that's usually me :-p). So, what if you use the C2D with "only" a CPU boost, memory running stock speed?
Thanks
Message edited by PeterHighlander on 07-20-2007 at 03:12:53 PM
| carod wrote : I'd love to see an article where multiple same model cpu's are overclocked, preferably all sourced from different different seller's to make it really random. |
That's a helluva idea. I'll see if we can make that happen.
I"m seeing Intel cpu's overclock less and less..
their binning down chips well and prevent the end user from getting significant gains.
that said, the E6320 I had was @ 1.375v on air; its the lowest possible voltage I could run at most stable.
intel could be a digital-crack dealer... get the enthusiast hooked on the "pure" stuff: high oc'ing chips. Then start "cutting" it with more and more sugar, err... binning until you are no longer getting anywhere close to the high you had at the start... but you can't stop buying it. They have you hooked for life.
lol.
Reply to sojrner
I didn't see that anyone mentioned this already so...
How about instead of 1 8800 GTX 640 card you get 2 8800 GTS 320 w/SLI. I have seen/read detailed benchmarks that show 2xSLI 8800 GTS cards significantly outperform 1 8800 GTX card, and for about the same cost.
how do you figure?
a decent 775 SLI board is around $149 cheapest....
The cheapest 8800GTS 320 is $279.... so 279x2=558
not to mention you better plunk down an extra $30-50 for a better PSU
to run those things.
no, a budget build is one GPU - a single 8800gts 320 would suit just fine as most people on a budget arent playing beyond 1650x1080
these are probably some of the most poor articles ever at toms hardware
I mean the day 2 and day 3 articles.
toms suffers from lack of professionalism .
I can find in my own neighbourhood at least 10 guys with high end, and some low priced overclocking machines which are far superior to what toms did here.
this lack of professionalism is going on for about 1 year.
Message edited by DECENEU on 07-22-2007 at 11:15:39 AM
Got to hate it when peeps quote your post and you've made a typo, especially when you're too much of a gimp to know how to edit it
Personally i thought it was a decent article and much better than some i've seen on toms in the not too distant past, keep at it cleeve some of us do appreciate the effort
Carod
Message edited by carod on 07-22-2007 at 09:27:08 PM
There are 52 identified and unidentified users. To see the list of identified users, Click here.
Please mind
You are about to answer a thread that has been inactive for more than 6 months.
If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.
