My E7300.....

I mentioned here a while back that I was going to upgrade my E4500
[65nm/2.40GHz/2MB] @ 3.32GHz/1.375V for an E7300 [45nm/2.66GHz/3MB] and that
I was hoping to get 4GHz out of it.

Sadly I was aiming a bit high. After much non-POSTing, not loading Windows,
BSODs and Prime fails I found that it topped out at 3.6GHz but that it
needed 1.400V set in BIOS [CPU-Z reading 1.380V] to be Prime-stable at that
speed. I ran it like that for a week or so but decided that I'd see what I
could get with a lower vcore. I can't afford to fry a CPU and these new
hafnium gate processors are supposed to be sensitive to vcore around 1.4V
and higher over a relatively short time.

Long story short, I'm running at 333 FSB now instead of default 266, multi
locked to 10 x and vcore on default. Temps are very low (they always have
been, not sure that core temp or core temp plugin for MBM5 is reading this
45nm CPU right). vcore in Windows seems to sit around the mid 1.2V range.

So, for my 'upgrade' I've got about the same speed as I had with the E4500
only at lower voltage, lower temps, an extra MB of L2 and a slightly more
efficient processor. (OCed E4500 gave me 3180/7128 Whetstone/Dhrystone
whereas the E7300 at about the same speed is giving me 3469/7179.) I'm
thinking that it's probably a bit cheaper to run too.

Not the big boost I was hoping for really. I've advertised it on the NZ
version of ebay at a $20 loss and, if it sells, I'll think about an E8400.
Maybe. If I sold the E7300 and squeezed my budget I could possibly get a
Q6600 but as my PC is running 24/7, that would cost me quite a bit more in
electricity costs. Also I think it would push the limits of my Tt
MiniTyphoon. Q9300s, while a lot more efficient, are still quite expensive
here.

Maybe I'll just stick with the E7300 @ 3.33GHz and default vcore. It's not
*that* bad a result I guess, 120% OC with no vcore boost. However, as a
dyed-in-the-wool OCer I feel a little unfulfilled. I think maybe the E4500
spoiled me. I do like a 150%+ OC.

I've taken note of those new E1200 / E1400 Celerons, 1.6GHz / 2GHz, 65nm
dual core CPUs for under NZ$100. Shame the L2 is so low. 512 KB. Then again,
I've heard of 200% OCs with the E1200 and it's said that the result is due
in part to the smaller L2 making them more OCable.

Cheers,
--
Shaun.

[Other stuff]
Asus P5K-E/WiFi-AP, BIOS ver. 1013
AcBel R8 700W PSU
2 x 1GB Transcend DDR2-800
Thermaltake MiniTyphoon.
Leadtek 7800GT.
2 x Seagate 500GB 7200.11 SATA II HDDs
2 x Seagate 320GB 7200.10 SATA II HDDs
Pioneer DVR-212 SATA DVD-RW
Windows XP Pro SP3.

"~misfit~" <> wrote in message
news:48aec093$...
>I mentioned here a while back that I was going to upgrade my E4500
>[65nm/2.40GHz/2MB] @ 3.32GHz/1.375V for an E7300 [45nm/2.66GHz/3MB] and
>that I was hoping to get 4GHz out of it.
>
> Sadly I was aiming a bit high. After much non-POSTing, not loading
> Windows, BSODs and Prime fails I found that it topped out at 3.6GHz but
> that it needed 1.400V set in BIOS [CPU-Z reading 1.380V] to be
> Prime-stable at that speed. I ran it like that for a week or so but
> decided that I'd see what I could get with a lower vcore. I can't afford
> to fry a CPU and these new hafnium gate processors are supposed to be
> sensitive to vcore around 1.4V and higher over a relatively short time.
>
> Long story short, I'm running at 333 FSB now instead of default 266, multi
> locked to 10 x and vcore on default. Temps are very low (they always have
> been, not sure that core temp or core temp plugin for MBM5 is reading this
> 45nm CPU right). vcore in Windows seems to sit around the mid 1.2V range.
>
> So, for my 'upgrade' I've got about the same speed as I had with the E4500
> only at lower voltage, lower temps, an extra MB of L2 and a slightly more
> efficient processor. (OCed E4500 gave me 3180/7128 Whetstone/Dhrystone
> whereas the E7300 at about the same speed is giving me 3469/7179.) I'm
> thinking that it's probably a bit cheaper to run too.
>
> Not the big boost I was hoping for really. I've advertised it on the NZ
> version of ebay at a $20 loss and, if it sells, I'll think about an E8400.
> Maybe. If I sold the E7300 and squeezed my budget I could possibly get a
> Q6600 but as my PC is running 24/7, that would cost me quite a bit more in
> electricity costs. Also I think it would push the limits of my Tt
> MiniTyphoon. Q9300s, while a lot more efficient, are still quite expensive
> here.
>
> Maybe I'll just stick with the E7300 @ 3.33GHz and default vcore. It's not
> *that* bad a result I guess, 120% OC with no vcore boost. However, as a
> dyed-in-the-wool OCer I feel a little unfulfilled. I think maybe the E4500
> spoiled me. I do like a 150%+ OC.
>
> I've taken note of those new E1200 / E1400 Celerons, 1.6GHz / 2GHz, 65nm
> dual core CPUs for under NZ$100. Shame the L2 is so low. 512 KB. Then
> again, I've heard of 200% OCs with the E1200 and it's said that the result
> is due in part to the smaller L2 making them more OCable.
>
> Cheers,
> --
> Shaun.
>

The Q6600, at least with my experience, is not going to get you any more as
far as raw Mhz is concerned......and not a lot more in performance other
than benchmarks like 3DMark. I am going to keep the Q6600 until next Spring
and probably look at a Nehalem. That will be about 2yrs which is my normal
upgrade cycle over the last 20yrs or so....:-). I am not sure if I will even
update video since my two 8800GTX cards are still pretty good performers in
SLI even compared to the newer Nvidia and ATI offerings. Power savings is
not a big deal and I doubt you would even notice it at all from an E4500 to
a Q6600. A very few pennies/mo would be the max. I added an EM64t @ 3.6Ghz
system as a 24/7 file server and can't see any difference in my monthly bill
that I can attribute to it or more AC running in the summer.........:-).

Ed

Somewhere on teh intarweb "Ed Medlin" typed:
> "~misfit~" <> wrote in message
> news:48aec093$...
>> I mentioned here a while back that I was going to upgrade my E4500
>> [65nm/2.40GHz/2MB] @ 3.32GHz/1.375V for an E7300 [45nm/2.66GHz/3MB]
>> and that I was hoping to get 4GHz out of it.
>>
>> Sadly I was aiming a bit high. After much non-POSTing, not loading
>> Windows, BSODs and Prime fails I found that it topped out at 3.6GHz
>> but that it needed 1.400V set in BIOS [CPU-Z reading 1.380V] to be
>> Prime-stable at that speed. I ran it like that for a week or so but
>> decided that I'd see what I could get with a lower vcore. I can't
>> afford to fry a CPU and these new hafnium gate processors are
>> supposed to be sensitive to vcore around 1.4V and higher over a
>> relatively short time. Long story short, I'm running at 333 FSB now
>> instead of default 266,
>> multi locked to 10 x and vcore on default. Temps are very low (they
>> always have been, not sure that core temp or core temp plugin for
>> MBM5 is reading this 45nm CPU right). vcore in Windows seems to sit
>> around the mid 1.2V range. So, for my 'upgrade' I've got about the same
>> speed as I had with the
>> E4500 only at lower voltage, lower temps, an extra MB of L2 and a
>> slightly more efficient processor. (OCed E4500 gave me 3180/7128
>> Whetstone/Dhrystone whereas the E7300 at about the same speed is
>> giving me 3469/7179.) I'm thinking that it's probably a bit cheaper
>> to run too. Not the big boost I was hoping for really. I've advertised it
>> on the
>> NZ version of ebay at a $20 loss and, if it sells, I'll think about
>> an E8400. Maybe. If I sold the E7300 and squeezed my budget I could
>> possibly get a Q6600 but as my PC is running 24/7, that would cost
>> me quite a bit more in electricity costs. Also I think it would push
>> the limits of my Tt MiniTyphoon. Q9300s, while a lot more efficient,
>> are still quite expensive here.
>>
>> Maybe I'll just stick with the E7300 @ 3.33GHz and default vcore.
>> It's not *that* bad a result I guess, 120% OC with no vcore boost.
>> However, as a dyed-in-the-wool OCer I feel a little unfulfilled. I
>> think maybe the E4500 spoiled me. I do like a 150%+ OC.
>>
>> I've taken note of those new E1200 / E1400 Celerons, 1.6GHz / 2GHz,
>> 65nm dual core CPUs for under NZ$100. Shame the L2 is so low. 512
>> KB. Then again, I've heard of 200% OCs with the E1200 and it's said
>> that the result is due in part to the smaller L2 making them more
>> OCable.
>
> The Q6600, at least with my experience, is not going to get you any
> more as far as raw Mhz is concerned......and not a lot more in
> performance other than benchmarks like 3DMark. I am going to keep the
> Q6600 until next Spring and probably look at a Nehalem. That will be
> about 2yrs which is my normal upgrade cycle over the last 20yrs or
> so....:-). I am not sure if I will even update video since my two
> 8800GTX cards are still pretty good performers in SLI even compared
> to the newer Nvidia and ATI offerings. Power savings is not a big
> deal and I doubt you would even notice it at all from an E4500 to a
> Q6600. A very few pennies/mo would be the max. I added an EM64t @
> 3.6Ghz system as a 24/7 file server and can't see any difference in
> my monthly bill that I can attribute to it or more AC running in the
> summer.........:-).

Hi Ed.

Yeah, those 8800GTX's should last you a while. I think that you're right
about the Q6600 not getting me more raw MHz, I had a play with one in a
system I built for a friend. Similar to mine only he went for the
P5K-Premium (one of those guys that has to go one better) and the Tt
Typhoon. It didn't clock as high as the E4500 and got quite hot. Heat =
power consumption and, here in NZ we don't have nuclear power and have a
moritorium on new fossil-fuel power stations so electricity is getting
expensive.

Add to that I'm on a (low) fixed income, an invalid's benefit
(OCing/hardware and teh intarweb are pretty much my only hobbies/use of
discretionary spending and that's often stretching it) and the fact that my
PC with the E4500 was costing me ~$10/week to run and you'll see why power
consumption is important to me. It dictates what I can eat each week. ;-)
Factor in the fact that I'm running SETI on a 50% cycle and two extra cores
would cost me a bit. (I suppose I could drop it to a 30% cycle in that case.
I ran a wattage meter on my machine/E4500 and found that running SETI @ 50%
only cost me an extra 15W above idle. Running at 100% cost me idle + 45W)

I used to have a two year turn over of computers too but since my injury and
subsequent income drop it seems to be panning out at around 5 years. That's
how long I kept my last machine, an XP2500+ Barton OCed to 2.2GHz. I'm not
that demanding of machines, the Barton and ti4200 would probably still do
everything I want, just with the odd second or two's pause here and there
(and it's not like I'm in a hurry really <g>).

So basically my aim pre-Nehalem was to build a machine that would be good
for 5 years with maximum bang-for-buck and some satisfaction to feed my OC
addiction. The reason I'm considering possibly changing CPUs at the moment
is I can probably still get ~85% of the cost of the E7300 second-hand if I
decide to sell. (I have a 100% positive rating on 'Trademe', our version of
ebay. http://www.trademe.co.nz/Members/Fee...x?member=33990 )

I haven't had any nibbles on my listing for the E7300 yet. I've pretty much
decided that I'll run two week-long auctions and let that decide if I'm
sticking with it or not. I put the E4500 into an Asus P5PE-VM,
http://www.asus.com/products.aspx?l1...76&modelmenu=1 an
odd-ball mATX mobo that I picked up for under NZ$100, runs 65nm dual cores
and gives me a capable back-up machine. It allows me to re-use my 2 x 1GB
DDR and my AGP card. Alas, it has no OCing ability and the BSEL mod that I
did had it booting at 266 FSB / 2.93GHz for a while Prime-stable but stopped
working and dropped back to 200 / 2.2GHz. I guess the de-fogger paint that I
used to make the connection either wore through (humidity? thermal cycling?)
or cracked.

Cheers,
--
Shaun.

DISCLAIMER: If you find a posting or message from me
offensive, inappropriate, or disruptive, please ignore it.
If you don't know how to ignore a posting, complain to
me and I will be only too happy to demonstrate... ;-)

Somewhere on teh intarweb "Bill" typed:
> In article <48aec093$>,
> says...


[snip]

Hi Bill,

> I know that this is not what you want to hear, but since you got
> similar results from 2 different processors maybe it's your mother
> board or memory that's holding you back.

I don't think that it is. The E4500 was running at 414MHz FSB (x 8) for the
OC I settled on (synchronous with RAM...) whereas the E7300 wouldn't do 10 x
400 but would do 8 x 450 or 9 x 400.

I've tried various combos of FSB and multi with both CPUs and the max
frequency for each was around the same regardless of how it was reached
(3.32 GHz for the E4500 and 3.62 GHz for the E7300). The mobo is stable to
at least 500MHz FSB and the RAM will loop Memtest all day at 450MHz / DDR900
with default vdimm.

The only reason that I got "similar results" is because I throttled the
E7300 back to a speed that it would do with default vcore as I can't afford
to 'lose' the E7300 and there's talk of them not liking long-term vcore near
or above 1.4V.

> The X48 chipsets are
> supposed to be a little better at OCing than the P35.

Only marginally, (if that) from what I've read.

> Then again, seeing the price of X48 motherboards, I'd probably stick
> with the E7300 and save my money for when Nehalem comes out.

I'll be still using 775 when Nehalem's been around for a few years
unfortunately. I can only swing one major purchase every five years or so
lately, hence the care with which I chose my mobo.

Cheers,
--
Shaun.

DISCLAIMER: If you find a posting or message from me
offensive, inappropriate, or disruptive, please ignore it.
If you don't know how to ignore a posting, complain to
me and I will be only too happy to demonstrate... ;-)

~misfit~ wrote:

> Long story short, I'm running at 333 FSB now instead of default 266, multi
> locked to 10 x and vcore on default. Temps are very low (they always have
> been, not sure that core temp or core temp plugin for MBM5 is reading this 45nm CPU right). vcore in Windows seems to
> sit around the mid 1.2V range.

I'm going to go ahead and ask you to try a 1333 BSEL mod now on that puppy!
http://www.ocforums.com/attachment.p...1&d=1204094697
Easy, no paint required.

Somewhere on teh intarweb "Fishface" typed:
> ~misfit~ wrote:
>
>> Long story short, I'm running at 333 FSB now instead of default 266,
>> multi locked to 10 x and vcore on default. Temps are very low (they
>> always
>> have been, not sure that core temp or core temp plugin for MBM5 is
>> reading this 45nm CPU right). vcore in Windows seems to sit around
>> the mid 1.2V range.
>
> I'm going to go ahead and ask you to try a 1333 BSEL mod now on that
> puppy!
> http://www.ocforums.com/attachment.p...1&d=1204094697
> Easy, no paint required.

Hehee! That does look simple! For a 1066 BSEL I had to cover the next land
up from that one and then run conductive material up from that one onto the
covered land above it.

Do you really want me to try it? I don't need to on my mobo as it has such
excellent OCing options in BIOS. However, if you'd like me to try it I will.
It's not like I haven't got about 4 tubes of different TIMs.

It's a shame that the mobo my E4500 is now in doesn't support 1333 FSB (or
vcore adjustment) or I'd try that on the E4500.

Cheers,
--
Shaun.

DISCLAIMER: If you find a posting or message from me
offensive, inappropriate, or disruptive, please ignore it.
If you don't know how to ignore a posting, complain to
me and I will be only too happy to demonstrate... ;-)

~misfit~ wrote:

> Do you really want me to try it? I don't need to on my mobo as it has such
> excellent OCing options in BIOS. However, if you'd like me to try it I will.
> It's not like I haven't got about 4 tubes of different TIMs.

They are saying it can cause it to boot with a different strap or something
and enable it to go higher on some boards. But since you were able to run
8 x 450, though, I don't know. Probably not? What the heck is Intel doing
to the low FSB parts to limit them?!

An interesting test would be to try each core separately. Maybe only one
is not up to snuff?

Somewhere on teh intarweb "Fishface" typed:
> ~misfit~ wrote:
>
>> Do you really want me to try it? I don't need to on my mobo as it
>> has such excellent OCing options in BIOS. However, if you'd like me
>> to try it I will. It's not like I haven't got about 4 tubes of
>> different TIMs.
>
> They are saying it can cause it to boot with a different strap or
> something and enable it to go higher on some boards.

Yes, it would. However my BIOS allows me to set the strap manually and
choose between 200, 266, 333 and 400. It's not dependant on the CPU to set
it.

> But since you
> were able to run 8 x 450, though, I don't know. Probably not?

No, there is truth in the 'strap' limiting max FSB. I think it's to do with
latency to NB. (<<< Semi-informed speculation, take with a grain of salt.)

> What the heck is Intel doing to the low FSB parts to limit them?!

See above. The E4500 (200 FSB stock) that I ran on a 450 FSB was done with
the strap set to 400.

> An interesting test would be to try each core separately. Maybe only
> one is not up to snuff?

Nah, I'm pretty sure it's a chipset and latency issue, nothing to do with
the CPU as such. The higher the strap the higher the CPU - NB latency is set
making it less likely to encounter errors with high FSB speeds. It is in
Intel thing but to do with chipset rather than CPU.

Cheers,
--
Shaun.

DISCLAIMER: If you find a posting or message from me
offensive, inappropriate, or disruptive, please ignore it.
If you don't know how to ignore a posting, complain to
me and I will be only too happy to demonstrate... ;-)

Somewhere on teh intarweb "Phil Weldon" typed:
> '~misfit~' wrote, in part:
>> Nah, I'm pretty sure it's a chipset and latency issue, nothing to do
>> with the CPU as such. The higher the strap the higher the CPU - NB
>> latency is set making it less likely to encounter errors with high
>> FSB speeds. It is in Intel thing but to do with chipset rather than
>> CPU.
> _____
>
> It must be a chipset feature; with my nVidia 680i chipset motherboard
> I can set the FrontSide Bus frequency to any integer (within reason)
> and the memory clock to any integer (within reason); some memory
> clock settings will not 'take', and the integer set will adjust by a
> few MHz. 'Course I have only an E4300 that tops out at about 3.3 GHz
> no matter what the multiplier.

Yep, now that you mention it I've heard it mentioned in relation to Intel
chipsets, mainly the P35 (which is no surprise as it's what I have so that's
what I've mostly been reading about).

> I see the run up to 'Nehalem' is beginning to affect prices;
> newegg.com is offering the Q9550 for less than $325 US.

<checks watch> Mother of God! Is it *that* time already??

I can feel myself falling off the (cutting) edge! It was fun while it
lasted. I'll probably disappear from the hardware and overclocking groups
again soon, as I did last when Socket A became old hat and nobody wanted to
know what I knew anymore...

I'll just be lurking in the shadows, reading about the fantastic new stuff
and waiting in case someone wants help with obsolete gear. <g>

Cheers,
--
Shaun.

DISCLAIMER: If you find a posting or message from me
offensive, inappropriate, or disruptive, please ignore it.
If you don't know how to ignore a posting, complain to
me and I will be only too happy to demonstrate... ;-)

Somewhere on teh intarweb "~misfit~" typed:
> Somewhere on teh intarweb "Fishface" typed:
>> ~misfit~ wrote:
>>
>>> Do you really want me to try it? I don't need to on my mobo as it
>>> has such excellent OCing options in BIOS. However, if you'd like me
>>> to try it I will. It's not like I haven't got about 4 tubes of
>>> different TIMs.
>>
>> They are saying it can cause it to boot with a different strap or
>> something and enable it to go higher on some boards.
>
> Yes, it would. However my BIOS allows me to set the strap manually and
> choose between 200, 266, 333 and 400. It's not dependant on the CPU
> to set it.
>
>> But since you
>> were able to run 8 x 450, though, I don't know. Probably not?
>
> No, there is truth in the 'strap' limiting max FSB. I think it's to
> do with latency to NB. (<<< Semi-informed speculation, take with a
> grain of salt.)
>> What the heck is Intel doing to the low FSB parts to limit them?!
>
> See above. The E4500 (200 FSB stock) that I ran on a 450 FSB was done
> with the strap set to 400.
>
>> An interesting test would be to try each core separately. Maybe only
>> one is not up to snuff?
>
> Nah, I'm pretty sure it's a chipset and latency issue, nothing to do
> with the CPU as such. The higher the strap the higher the CPU - NB
> latency is set making it less likely to encounter errors with high
> FSB speeds. It is in Intel thing but to do with chipset rather than
> CPU.

..... and just for the hell of it I've just changed my settings from....

.....10 x, 333 FSB strap, 333 FSB, DDR2-800 and default vcore to...

.... 7.5 x, 400 FSB strap, 450 FSB, DDR2-900 and still on default vcore.

(I tried 7 x 500 but it didn't like it, I think because the mimimum RAM
speed on this board is = to FSB and I only have Transcend RAM rated for
DDR2-800. 450 FSB is it for me unless I buy faster RAM.)

I've only increased the CPU speed by a little, from 3.33 to 3.37 but the
vcore reading by CPU-Z in Windows has gone up from 1.28 to 1.34V (I'm not
sure I like how these modern fandangled CPUs change their own vcore, I've
disabled all the Intel CPU power-saving stuff in BIOS...)

The CPU benches a bit better at this setting (probably why the vcore has
gone up, it's doing more work) and I'm sure the RAM is better at 450 than it
was at 400... Also CPU FSB and RAM are now synchronous.

I can't seem to stop playing.... It's a good job I have a computer to do it
with and not just myself. I do like my hardware. ;-)
--
Shaun.

DISCLAIMER: If you find a posting or message from me
offensive, inappropriate, or disruptive, please ignore it.
If you don't know how to ignore a posting, complain to
me and I will be only too happy to demonstrate... ;-)