BloodRAGE PCI-E and Bclk Mod

[saaya]

but hay Saaya how did you miss that one if you know of this mod why you not say

i didnt know which resistor it was on BR... i heard of this mod for other boards, i told our engineers a couple of weeks before i left but they were not interested in getting higher pciE clocks working...
i asked them several times and they always said they think its not important...

mescalamba, if you can get higher bclocks without the mod, once you do the mod you should be flying im not sure how exactly this works but i suspect the IOH becomes unstable when qpi and pciE clocks are too far apart.

funny actually, cause it was the same with the 875 chipset and agp clocks and other older intel chipsets
increasing agp/pci clocks and increasing those bus voltages made small to notable diferences in the past, so im not surprised it helps here as well... though for some it doesnt seem to help at all, for others it helps a lot...

on 875 the diference was incredible, i remember hipro modding his asus or abit 875 board to increase agp volts and clocks and got insane fsb clocks on socket 478...

ahh there it is, abit it was... and he got 400fsb on a socket 478 system... in 2003!
http://www.xtremesystems.org/forums/...ad.php?t=24520

[saaya]

Utnorris, i dont think its a board issue... and if you want the mod done, just ask nicely and foxconns rma center might actually do it, but dont try to force the issue and claim you deserve a modded board from them, thats a bit how your posts sound

[Utnorris]

If that is how it comes across that was not my intention. I was looking for a better pic to see which resistor it was. My main concern has been the issue with the voltage fluctuating, it's actually a little scary since it can jump back to the original default voltage which would sky rocket my vcore. To get 4.2Ghz stable I have to add about .05v from the 1.23v default, to get the same result from 1.13v I have to add .15v vcore. At the 4.2Ghz it's not a real issue, but jumping up to 4.4Ghz I have to be at 1.4v vcore on my chip, that's a .17v jump from the 1.23v or .27v from the 1.13v vcore. If it jumps back up to the default 1.23v default vcore that puts my vcore up to 1.5v which I would rather not run, especially on a 24/7 basis. This mod wouldn't help my situation either way. I am hoping I can get the G25 bios to fix it, but since I am not seeing anyone else with the same issue it leads me to believe it is a board issue. Again, wasn't trying to get the tech support to do the mod, just trying to fix my issue with the default vcore voltage.

[hipro5]

Have you measured the PCI-E frequency with a frequency counter to be sure that it is 120MHz OR just of what you see in setfsb?

I'm asking you this because the datasheet of the PLL Clock generetor IC writes down that whatever you pick as for PCI-E frequency, it locks it at 100MHz...


Bios engineers of Bloodrage mobo SUCK BIG TIME....
Why?
In bios G22 and G25 (haven't tested the other's) the VCore and the VQPI are OPPOSITE...
When you change the VCore into bios, ACTUALY you change the VQPI and when you change the VQPI, it changes the VCore...

Go someone and get a multimeter and measure the VCore...
Go into bios and change it...
You'll read the SAME VCore after save and exit, bla, bla, bla...
Now go and change the VQPI and you will see that the VCore is changing...

I don't know IF this is involved with the pre-released BloodRage mobo or not, I just say what I see and measure because I was about to burn down my CPU by thinking of feeding it with more VCore, though I gave it OVER 1.65 VQPI without knowing it...

[Lu(ky]

Question does this apply to all Bloodrage boards, or certain version that came out? I never new the board had a 220 and above problem, which really sucks. I have mine sitting here new never used for that past 2 months and now I am thinking maybe send it back for a evga..

[Lord Trident]

Hipro5 i will check and if it true i get the bios people to fix
this is the first i know of this problem and no one has said before.

Lu(ky why does this sucks have you tried it as a lot of people can not get 220 to start with
yes i know there a few of you that have and this is why am trying to get them fixed
the pci-e freq fix helps the bckl but as far as i know we still fixing the bckl problem as you should be able to do this with out pci-e freq
i hope to have more information on this soon.

thanks

[Mescalamba]

Question does this apply to all Bloodrage boards, or certain version that came out? I never new the board had a 220 and above problem, which really sucks. I have mine sitting here new never used for that past 2 months and now I am thinking maybe send it back for a evga..

Not even eVGA can make 220 easy.. it depends on your CPU and piece of mobo, if you get bad piece of CPU or mobo.. you'll be happy to make stable 200.. I've seen this before on quite a lot i7 boards. 220 is quite enough, if you don't plan to put your CPU on stage cooling.

[Lu(ky]

Not even eVGA can make 220 easy.. it depends on your CPU and piece of mobo, if you get bad piece of CPU or mobo.. you'll be happy to make stable 200.. I've seen this before on quite a lot i7 boards. 220 is quite enough, if you don't plan to put your CPU on stage cooling.

Well I do have a 2 good DO I just got, I was planning on this being my show board with case mod I been messing with. I have no plans on selling it.. I will try in the next couple of days with this DO Batch like this chip --> H E R E

[Mescalamba]

Well I do have a 2 good DO I just got, I was planning on this being my show board with case mod I been messing with. I have no plans on selling it.. I will try in the next couple of days with this DO Batch like this chip --> H E R E

SLBEJ .. hm, I think these are quite good, you should be able to do 4,2 - 4,4 without much problems on BR.. more will need better cooling. 220x21 = 4620 - means probably no go without water cooling or TEC or both.. but ofc depends how much voltage your CPU needs for that.

Having more than 220 BCLK is nice, but with i7 920 and air cooling, it doesn't have much meaning (except fact that higher your FSB = faster it is).

[r1ch]

[B]Have you measured the PCI-E frequency with a frequency counter to be sure that it is 120MHz OR just of what you see in setfsb?

Sorry George I don't have one of those - I can do some benchmark's though

3D03, 9800GX2@760/1800/1155, 920@220x12=2640
PCI-E@100 = 68671
PCI-E@109 = 68844
PCI-E@119 = 69198

http://www.benchtec.co.uk/images/rich/bloodrage/100.jpg
http://www.benchtec.co.uk/images/rich/bloodrage/109.jpg
http://www.benchtec.co.uk/images/rich/bloodrage/119.jpg

I think it's working!

Bios engineers of Bloodrage mobo SUCK BIG TIME....
Why?
In bios G22 and G25 (haven't tested the other's) the VCore and the VQPI are OPPOSITE...
When you change the VCore into bios, ACTUALY you change the VQPI and when you change the VQPI, it changes the VCore... ]

I've never come across this myself, and I have a pre-release board like you. I always have a DMM on hand with this board though. From the testing I've done it can fluctuate the vcore based on how many cores/threads you have enabled, and what CPU multi you have selected (although I think the CPU multi is meant to have improved with G25 BIOS as they changed the VID tables.

Lu(ky, it may just be my board that refused to go over exactly 220, but other have run into problems around there, i.e in the 220/221/222 range. Some have been lucky and got to 225 and after the G25 BIOS I got to 229.

http://valid.canardpc.com/show_oc.php?id=576031

I 'only' have a C0 as well, so I don't know how D0's play with it but they can't be worse - i'd give the bloodrage a go, and if you need more bclk then see if you can get another board. I don't think you will - 4.6Ghz+ is difficult with ANY board.

[saaya]

(except fact that higher your FSB = faster it is).

for fsb thats true cause it affects the chipset clocks, but for bclock? if you increase the multipliers and reduce bclock to have the same uncore, core and memory clocks, the performance will be the same... at least from what i tested?

4.6Ghz+ is difficult with ANY board.

on air

george,
so the early board i sent you has vtt and vcore swapped with the newer bios versions? thats odd...

carl, can you tell the bios guys about this? thx

[systemviper]

Bios engineers of Bloodrage mobo SUCK BIG TIME....
Why?
In bios G22 and G25 (haven't tested the other's) the VCore and the VQPI are OPPOSITE...
When you change the VCore into bios, ACTUALY you change the VQPI and when you change the VQPI, it changes the VCore...

Go someone and get a multimeter and measure the VCore...
Go into bios and change it...
You'll read the SAME VCore after save and exit, bla, bla, bla...
Now go and change the VQPI and you will see that the VCore is changing...

[r1ch]

Do you have a bloodrage with the same problem systemviper?

I'm not doubting what George is saying, but if this problem is limited to his pre-release, engineering sample board then I don't think it's a big deal. I'd hope Foxconn will sort George out with a BIOS, or a mod that fixes things for him.

Like I said, I have a pre-release board as well and I have never seen this. I don't think I would be benching at 4.5-4.6Ghz with stock vcore and 1.5vqpi

[Mescalamba]

Do you have a bloodrage with the same problem systemviper?

I'm not doubting what George is saying, but if this problem is limited to his pre-release, engineering sample board then I don't think it's a big deal. I'd hope Foxconn will sort George out with a BIOS, or a mod that fixes things for him.

Like I said, I have a pre-release board as well and I have never seen this. I don't think I would be benching at 4.5-4.6Ghz with stock vcore and 1.5vqpi

I hope he doesn't cause I have now his motherboard. Well, time to test.. lol..

[Mescalamba]

for fsb thats true cause it affects the chipset clocks, but for bclock? if you increase the multipliers and reduce bclock to have the same uncore, core and memory clocks, the performance will be the same... at least from what i tested?

on air

george,
so the early board i sent you has vtt and vcore swapped with the newer bios versions? thats odd...

carl, can you tell the bios guys about this? thx

Ah.. so.. no point in having 4GHz via 200x20 instead of 190x21?

Interesting.. That probably explains how I can get so much bandwith even with quite crappy rams.. (tough, latency not good.. 50ns).

[systemviper]

Do you have a bloodrage with the same problem systemviper?

I'm not doubting what George is saying, but if this problem is limited to his pre-release, engineering sample board then I don't think it's a big deal. I'd hope Foxconn will sort George out with a BIOS, or a mod that fixes things for him.

Like I said, I have a pre-release board as well and I have never seen this. I don't think I would be benching at 4.5-4.6Ghz with stock vcore and 1.5vqpi

i'l have to break out the vmeter and find out, but i was just thinking how wak that would be...

[r1ch]

I hope he doesn't cause I have now his motherboard. Well, time to test.. lol..

i'l have to break out the vmeter and find out, but i was just thinking how wak that would be...

You have another bloodrage?

Certainly is odd. Only hipro and Foxconn can tell us any more

[Mescalamba]

You have another bloodrage?

Certainly is odd. Only hipro and Foxconn can tell us any more

I have two now. One from some guy in UK and second from Systemviper.

[Mescalamba]

Btw. I like this mod but.. will we be able to have 120 PCI-E without this mod somewhere in future? Or its only way to get more PCi-E freq?

[Lord Trident]

think that could be the limit of pci-e feq but again it could be any card that on the bus holding back on 119
so you may find you can go higher by turning off almost every bit
like lans as they are pci-e chips raid chips and graphic card could hold back
but i will find out if we can get a bit more or not on the pci-e but why you want 120, may i ask and no, you need the mod to get. the bios can not fix
it a wrong track link on the north chip that what the mod is doing

it should have been picked up in the early days but no one seem to ask or try pci-e freq
even i missed that but then again i do tv card testing and i know a lot of tv cards on pci-e get upset about the bus over 104 lol
so i dont touch my self

[Mescalamba]

think that could be the limit of pci-e feq but again it could be any card that on the bus holding back on 119
so you may find you can go higher by turning off almost every bit
like lans as they are pci-e chips raid chips and graphic card could hold back
but i will find out if we can get a bit more or not on the pci-e but why you want 120, may i ask and no, you need the mod to get. the bios can not fix
it a wrong track link on the north chip that what the mod is doing

it should have been picked up in the early days but no one seem to ask or try pci-e freq
even i missed that but then again i do tv card testing and i know a lot of tv cards on pci-e get upset about the bus over 104 lol
so i dont touch my self

120 is limit for usability usually .. after 120 lots of graphic cards or audio etc. , even SATA HHD refuse to work.. thats why 120.

Hm.. so I need that mod, well.. time to buy some solder-ware.

For myself, 110 - 115 is enough.

[r1ch]

I have two now. One from some guy in UK and second from Systemviper.

I was meaning systemviper

120 is limit for usability usually .. after 120 lots of graphic cards or audio etc. , even SATA HHD refuse to work.. thats why 120.

Hm.. so I need that mod, well.. time to buy some solder-ware.

For myself, 110 - 115 is enough.

I agree, I've not really tried past 120 because of my daily rig and SATA HDD corruption.

Practise (a lot) first, this is a very small and tricky mod.

[systemviper]

think that could be the limit of pci-e feq but again it could be any card that on the bus holding back on 119
so you may find you can go higher by turning off almost every bit
like lans as they are pci-e chips raid chips and graphic card could hold back
but i will find out if we can get a bit more or not on the pci-e but why you want 120, may i ask and no, you need the mod to get. the bios can not fix
it a wrong track link on the north chip that what the mod is doing

it should have been picked up in the early days but no one seem to ask or try pci-e freq
even i missed that but then again i do tv card testing and i know a lot of tv cards on pci-e get upset about the bus over 104 lol
so i dont touch my self

yea it seems that upping the pciE is th magic bullet on a few other boards too, like the gig board and i will try it on the gene too.

did i read earlier that the bios can only get you 103 and you need setfsb to go higher. or is that only with the hard mod. i try to avoid hard mods...

Can you post or IM you setings for the high bclocks. i want to make sure i have all the other settings on target also.

thanks for your effort, the BR seems to be responding nicely to this new info and upgrades

[Mescalamba]

yea it seems that upping the pciE is th magic bullet on a few other boards too, like the gig board and i will try it on the gene too.

did i read earlier that the bios can only get you 103 and you need setfsb to go higher. or is that only with the hard mod. i try to avoid hard mods...

Can you post or IM you setings for the high bclocks. i want to make sure i have all the other settings on target also.

thanks for your effort, the BR seems to be responding nicely to this new info and upgrades

Nope. Only hard mod, as Lord Trident said, theres missing circuit, they didn't intend to try other than default PCI-E freq.. so if I am correct, that circuit is still there, just need that resistor to work. I've looked on that.. huh its really really small.

[saaya]

carl, the engineers actually told me they use a seperate clock signal for the x58 pciE and the sb pciE, if thats true, then the onboard components should not be affected by raising the pciE clocks... im not sure if the sonar card clocks are related to those pciE though... but i dont think so.
i told them early in the beginning of making BR that pciE vga clocks should be independant from any other clock so it can be overclocked without causing raid or other issues... and they said they will do it, and later i checked and they said the clocks are independant...