Soldato
- Joined
- 23 May 2006
- Posts
- 8,340
That is absolutely not true. Sure if you take the p... It will have an effect but good cover absolutely is for personal belongings and accidental damage. Just don't claim every year on it.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
I broke it! | RTX 3090 Water block install gone wrong
- Thread starter TR1PL3D
- Start date
Soldato
- Joined
- 3 Oct 2007
- Posts
- 12,181
- Location
- London, UK
Sorry that is nonsense. Why do you add accidental damage to policies and state a maximum value of certain items? My PC is by far the most expensive item I have in the house so I stated its value on my policy. If you don't want to include such things you can probably save a few £. I've claimed on household insurance twice in 25 years. I think they have done ok out of me tbh.
Associate
- Joined
- 7 Apr 2006
- Posts
- 292
- Location
- USA
Ugh
I was hoping @Maldoror would reply with information about the mods he did to pass the 580W limit, but I guess he doesn't want to.
It would be really useful to both FE users and others but...welp...
I was hoping @Maldoror would reply with information about the mods he did to pass the 580W limit, but I guess he doesn't want to.
It would be really useful to both FE users and others but...welp...
how about you help the community out and tell/show us in pictures what you did, because everyone is stuck at 550-580w on their shuntmodded FE but you.
Associate
- Joined
- 7 Apr 2006
- Posts
- 292
- Location
- USA
Yeah that's me.
And I just bought a new soldering iron thanks to a very experienced solderer on Elmor's discord, a TS100, with a 24V power supply, its a 65W iron so it should make stacking shunts MUCH easier than the crappy "starter" iron I was using before.
I'm just waiting for my 60/40 solder to arrive before I redo the GPU Chip Power Shunt.
But I'd like to also do whatever "mod" is required to bypass that 580W limit.
But those tiny shunts...you cant just experiment on them like you can the big ones. They're too small to stack anything on, and sure, you could get MG 842AR paint on them, *IF* you use 3M high temp Kapton tape and carefully insulate EVERYTHING around them--I'm not even sure if Super 33+ tape would work because there's too many tiny parts around PCIE Slot power!!, but then...how would you reverse the mod? You would have to apply like a thin coat of paint and then wipe it with 100% alcohol somehow, with the entire area taped off...but the alcohol would just weaken the adhesive and then residue would get between those tiny SMD's....Yeah...have fun getting that residue out....I don't think anyone wants to spend an hour trying to clean stuff like that... Yeah there's more room around the junior 005 next to "GPU Chip Power"--I'll admit that. But PCIE Slot? uh...yeah...
That's why we need to know exactly which mods that guy did. Even if he doesn't know WHICH one made it work...at least telling us all the mods he 'attempted' gives us something to at least go on...
Anyway....
I am already 99% (not 100%, giving myself an out here) sure where this limit it's coming from in hwinfo64.
It's the limits that do NOT respond to shunt mods, as I mentioned above, several of the GPU Core NVVDD Output Power rails, that end up *higher* than Total Board Power, when all the main shunts are shunted.
I am NOT talking about GPU Chip Power itself. I already determined that GPU Chip Power is a sum of three other rails and I found them already:
GPU Chip Power=GPU Misc1 Input Power + GPU Misc3 Input Power + GPU Core NVVDD1 Input Power (sum).
This responds to shunting GPU Chip Power.
Note that this is an INPUT RAIL! The shunts mods affect the INPUT RAILS!!! PCIE Slot, MVDDC, SRC, 8 pin#1, #2, GPU Chip Power...these are input rails.
But there are other rails....and it's some of the OUTPUT rails which seem to be throwing a spanner into the works!
namely: GPU Core NVVDD Output Power (sum)= (Bottom-most) GPU Core NVVDD Output Power + GPU SRAM Output Power. I don't think SRAM output power is the culprit however.
It's that last output power that seems to ignore shunt mods (SRAM Output power seems linked to SRAM Input power, I don't "think" this is involved, but of course I don't know for sure).
This power rail also seems to affect TDP Normalized%. I already know that Normalized seems to read the 8 pins like this:
100% Normalized=150W 8 pin= throttle point
114% normalized=175W 8 pin= throttle point
(Note: the 8 pin limits seem to be the same as the SRC limits shown in the ampere bios editor: 150W default, 175W maximum).
100% normalized=235W GPU Chip Power limit
114% normalized=270W GPU Chip Power limit (it seems that GPU Chip Power can go up to 285W before throttling however---weird).
PCIE Slot limit: 69W 100%, 79W=114%. Yes I know the vBIOS says 86.2W, but it never gets there...
Anyway---back to GPU Core NVVDD Output Power (sum). OR GPU Core NVVDD Output Power.
Preliminary tests "seem" to show this throttling close to 336W at 100% Normalized TDP and around 370-380W at 114% normalized TDP. This is for the "summed" value.
Note that this can be WAY higher than TDP%, because TDP% is just the two 8 pins + PCIE slot power added together!
However I do NOT know if it is the "sum" rail or the independent GPU Core NVVDD Output Power (the one right above SRAM output power) by itself that has the limit attached to it.
This isn't exact and I am not 100% confident on these numbers. But there are simply no other power rails left.
I'm really not sure what trickery Malador did to bypass the limit.
If you guys have Superposition, you can do a "4k custom--extreme shaders" test. You will see that this "Sum" ends up exceeding 400W--hitting 450W! (if you have a shunted GPU) and you get a THICK permanent throttle bar throughout the test.
The output power by itself hits 300W, with Sram output power=150W...
Last edited:
I only log in here when gaming is quiet so this is the first time I've seen the message. I'll outline some of my adventures with the FE below and perhaps there's something in there you find useful. It's a more fussy card than any I've previously overclocked and my insight should probably be filed more under alchemy than science.
As a general proviso to what I'll say next, I'm into overclocking but only to further my in-game performance. There have been battles won and events recorded in Eve Online's history on the basis of, say, having tri-SLI TitanXs versus not having them. It's less of a factor today in that game but this kind of thing really happened. So I'm less concerned with power draw in 3dmark or Superposition compared to power draw in a 3000 man MMO fight with all details on, 12k resolution, etc.
Therefore I'd muse that it's possible that your card will also pull the same power in that specific instance, given that the 1100mV of an FE isn't likely to push 600w in Port Royal, and presently you can't solder an Elmor anywhere to change that (that I know of anyway?). We have seen some examples of extreme power draw in Path of Exile, for example, with the highest shadow setting and quirks like that.
Another thing I'd point out is that I only bench with the settings I game on. So I launch 3dmark through Steam, keep textures on 'Quality' in CP, etc. Doing a run like that just now on Port Royal, gives 15457 (https://www.3dmark.com/pr/841767 ) and I don't think it exceeded around 600w draw.
Which leads to the next thing, before I talk about modding, which is how power is measured.
I have two means: one is to run CPU/board on a different PSU, and then use my Asus Thor 1200 to monitor at least the 8pin connections (using its OLED display) - though of course this doesn't include the PCIE power.
I had to abandon this method because my shunt mod somehow stopped the Thor 1200 working properly - it randomly trips OCP on the card (despite being nowhere nearly total OCP, which is in fact way above 1200w on a Thor, and we're not close to 1200 anyway). I switched to using an EVGA 1600 T2 and this solved the problem completely (probably due to the analogue controller); I haven't had a crash in weeks on my 3090, even benching. I value stability above everything else (you crash at the wrong time in games like Eve, and you might lose more than the value of a 3090).
So with the 1600 T2 plugged into a wall monitor, my other method is to compare the load to an unmodded Titan RTX (my modded one messed with psus as well but that's a different story). I know the peak draw of the unmodded Titan is 320w. So let's say a given test pulls 550w total system power on the Titan, and at that moment the Titan is 100% load and showing 320w in AB, right up on its limit. If the 3090, in the same scene, is then pulling 800w from the wall, we can say that it's pulling 250W more than the Titan, so 320+250 = 570w in that specific example. Of course, we also have to acknowledge that the slightly higher fps of a given test under the 3090 might mean marginally more work for the CPU - however, I don't think that increase is likely to be more than 10-20w, and that's generally within the margin of error of the wall monitor anyway. Even though we're measuring from the wall, PSU efficiency doesn't play a major role in this since we are comparing like with like - arguable minor variance in PSU efficiency across its power curve might affect our results slightly, but again I'd argue this to be less than 10w.
As for modding:
Initially I shunted the 8pins and left the rest alone, as I used to on previous cards. As you'd now expect, this didn't produce great results - the power stayed around 400w, and SRC showed odd values.
Then I decided to shunt all the 5mohm resistors (by stacking with 5mohm) though this led me to be concerned about the PCIE current. To combat that at little, I'm using EVGA's old supplemental PCIE power plugs, to reduce the trace distance across the length of the board from the 24 pin to the PCIE slot, the journey along which I'd expect problems to occur if we're exceeding 100-120w draw from the PCIE.
This produced a (maximum) power draw that my above method estimates around 550-580w. It would tend to cap these kind of values even the the tdp target wasn't exceeded, which goes along with what you mentioned about hidden limits.
Around the same time, I had the problem that 8pin #1 was reading a lot higher than 8pin #2 (8pin 1 being the left one of the shunts when viewed from above with the display port facing left). This is a really fiddly place to solder and drove me mad trying to get it right, so I swapped that with a 3mohm, and for some reason also swapped the one below with a 3mohm (which on multimeter shows continuity). This helped balance them a bit but still somewhat odd results.
From somewhere I had some tiny 8mohm resistors in a smaller package than the usual ones, so I put these onto the tiny shunts nearest the core, relatively speaking (even though the size didn't match perfectly, making it messy). I'm not sure if this truly did anything productive or not and at the time I also changed the top left main shunt on the back (displayport facing right) with an 8mohm, for reason I can't remember (possibly red wine).
Anyway, around this time, PWR_SRC started sometimes reading 0 (although I have heard on here about that happening to others too) and an odd thing would happen often, where the card would just get stuck in 'IDLE' even under full load. I took screenshot the first time it happened (still on air cooling at that point):
http://rooksandkings.com/mini/idleload.jpg
So you can see the card in sections there is pinned at 1100mV, running full clock and load but stuck in performance mode 'IDLE' rather than vOP/ReL/PWR or TMP. In this mode, using the above methods, I measured it pulling 700w+ from the wall. As you can see on the timeline, I was puzzled enough by this that I started it up again and as you can see, it ramps back up to 1100mV and full load while staying in IDLE, not vOP/REL. It also holds the 'Effective' clock very close to the display clock in Tempspy.
Some other points: I can't run this card on an Asus Thor 1200. It'll trip after an hour or two. I've run it for weeks on an EVGA 1600 T2, no issue, 100% stable with 2200mhz on the core now that's on a loop. Sometimes it runs on vREL/OP but it does usually hold 1100mV.
Now, if you want to *really* go down the rabbit hole here's something absurd though I can't verify it: after explaining this to a friend, he was (for other reasons) repeatedly opening and closing HWinfo, Tempspy, afterburner, precision, and at at some point his unshunted FE got 'jammed' and stuck in idle under full load, where it appeared (alas he can't measure wattage but judging by a bench he ran) to be drawing well above 400w and holding 1100mV. After reboot it couldn't be repeated, no matter what he tried. Make what you will of that part - I have no idea on that one; I modded my card in the launch week for 3090 and can't test that. Still, part of me believes there is a way to trick the entire monitoring system of these cards (even unshunted) and essentially use the IDLE mode to lock 1100mV and draw whatever consequent power that entails.
I can assure you that I'm no soldering wizard but if I take the block off again (Corsair block - all I could source up to now) I'll take you a PCB picture of the mess. Normally I'd offer to do it just to indulge your curisosity but the stacked shunts initially made - I think - contact with the block, needing me to insulate the tops against a potential short and generally it took me 3 attempts to get the block on satisfactorily and I'm scared to take it back off except to change it to a better block (even a good loop is only giving me 35-40C or so).
I doubt any of this is particularly helpful because it's as much alchemy as overclocking. I've never had a board that has such odd power balancing and although I love the peformance generally, I can't pretend to properly understand it. If I can give you anything, it's my belief that there's some kind of way to trick this board using the 'IDLE' power mode and if I ever figure it out I'll certainly post it up.
Anyway, sorry for the delay and the long essay - I realise that 'adventures with a pcb' makes for very specific bedtime reading
I only log in here when gaming is quiet so this is the first time I've seen the message. I'll outline some of my adventures with the FE below and perhaps there's something in there you find useful. It's a more fussy card than any I've previously overclocked and my insight should probably be filed more under alchemy than science.
As a general proviso to what I'll say next, I'm into overclocking but only to further my in-game performance. There have been battles won and events recorded in Eve Online's history on the basis of, say, having tri-SLI TitanXs versus not having them. It's less of a factor today in that game but this kind of thing really happened. So I'm less concerned with power draw in 3dmark or Superposition compared to power draw in a 3000 man MMO fight with all details on, 12k resolution, etc.
Therefore I'd muse that it's possible that your card will also pull the same power in that specific instance, given that the 1100mV of an FE isn't likely to push 600w in Port Royal, and presently you can't solder an Elmor anywhere to change that (that I know of anyway?). We have seen some examples of extreme power draw in Path of Exile, for example, with the highest shadow setting and quirks like that.
Another thing I'd point out is that I only bench with the settings I game on. So I launch 3dmark through Steam, keep textures on 'Quality' in CP, etc. Doing a run like that just now on Port Royal, gives 15457 (https://www.3dmark.com/pr/841767 ) and I don't think it exceeded around 600w draw.
Which leads to the next thing, before I talk about modding, which is how power is measured.
I have two means: one is to run CPU/board on a different PSU, and then use my Asus Thor 1200 to monitor at least the 8pin connections (using its OLED display) - though of course this doesn't include the PCIE power.
I had to abandon this method because my shunt mod somehow stopped the Thor 1200 working properly - it randomly trips OCP on the card (despite being nowhere nearly total OCP, which is in fact way above 1200w on a Thor, and we're not close to 1200 anyway). I switched to using an EVGA 1600 T2 and this solved the problem completely (probably due to the analogue controller); I haven't had a crash in weeks on my 3090, even benching. I value stability above everything else (you crash at the wrong time in games like Eve, and you might lose more than the value of a 3090).
So with the 1600 T2 plugged into a wall monitor, my other method is to compare the load to an unmodded Titan RTX (my modded one messed with psus as well but that's a different story). I know the peak draw of the unmodded Titan is 320w. So let's say a given test pulls 550w total system power on the Titan, and at that moment the Titan is 100% load and showing 320w in AB, right up on its limit. If the 3090, in the same scene, is then pulling 800w from the wall, we can say that it's pulling 250W more than the Titan, so 320+250 = 570w in that specific example. Of course, we also have to acknowledge that the slightly higher fps of a given test under the 3090 might mean marginally more work for the CPU - however, I don't think that increase is likely to be more than 10-20w, and that's generally within the margin of error of the wall monitor anyway. Even though we're measuring from the wall, PSU efficiency doesn't play a major role in this since we are comparing like with like - arguable minor variance in PSU efficiency across its power curve might affect our results slightly, but again I'd argue this to be less than 10w.
As for modding:
Initially I shunted the 8pins and left the rest alone, as I used to on previous cards. As you'd now expect, this didn't produce great results - the power stayed around 400w, and SRC showed odd values.
Then I decided to shunt all the 5mohm resistors (by stacking with 5mohm) though this led me to be concerned about the PCIE current. To combat that at little, I'm using EVGA's old supplemental PCIE power plugs, to reduce the trace distance across the length of the board from the 24 pin to the PCIE slot, the journey along which I'd expect problems to occur if we're exceeding 100-120w draw from the PCIE.
This produced a (maximum) power draw that my above method estimates around 550-580w. It would tend to cap these kind of values even the the tdp target wasn't exceeded, which goes along with what you mentioned about hidden limits.
Around the same time, I had the problem that 8pin #1 was reading a lot higher than 8pin #2 (8pin 1 being the left one of the shunts when viewed from above with the display port facing left). This is a really fiddly place to solder and drove me mad trying to get it right, so I swapped that with a 3mohm, and for some reason also swapped the one below with a 3mohm (which on multimeter shows continuity). This helped balance them a bit but still somewhat odd results.
From somewhere I had some tiny 8mohm resistors in a smaller package than the usual ones, so I put these onto the tiny shunts nearest the core, relatively speaking (even though the size didn't match perfectly, making it messy). I'm not sure if this truly did anything productive or not and at the time I also changed the top left main shunt on the back (displayport facing right) with an 8mohm, for reason I can't remember (possibly red wine).
Anyway, around this time, PWR_SRC started sometimes reading 0 (although I have heard on here about that happening to others too) and an odd thing would happen often, where the card would just get stuck in 'IDLE' even under full load. I took screenshot the first time it happened (still on air cooling at that point):
http://rooksandkings.com/mini/idleload.jpg
So you can see the card in sections there is pinned at 1100mV, running full clock and load but stuck in performance mode 'IDLE' rather than vOP/ReL/PWR or TMP. In this mode, using the above methods, I measured it pulling 700w+ from the wall. As you can see on the timeline, I was puzzled enough by this that I started it up again and as you can see, it ramps back up to 1100mV and full load while staying in IDLE, not vOP/REL. It also holds the 'Effective' clock very close to the display clock in Tempspy.
Some other points: I can't run this card on an Asus Thor 1200. It'll trip after an hour or two. I've run it for weeks on an EVGA 1600 T2, no issue, 100% stable with 2200mhz on the core now that's on a loop. Sometimes it runs on vREL/OP but it does usually hold 1100mV.
Now, if you want to *really* go down the rabbit hole here's something absurd though I can't verify it: after explaining this to a friend, he was (for other reasons) repeatedly opening and closing HWinfo, Tempspy, afterburner, precision, and at at some point his unshunted FE got 'jammed' and stuck in idle under full load, where it appeared (alas he can't measure wattage but judging by a bench he ran) to be drawing well above 400w and holding 1100mV. After reboot it couldn't be repeated, no matter what he tried. Make what you will of that part - I have no idea on that one; I modded my card in the launch week for 3090 and can't test that. Still, part of me believes there is a way to trick the entire monitoring system of these cards (even unshunted) and essentially use the IDLE mode to lock 1100mV and draw whatever consequent power that entails.
I can assure you that I'm no soldering wizard but if I take the block off again (Corsair block - all I could source up to now) I'll take you a PCB picture of the mess. Normally I'd offer to do it just to indulge your curisosity but the stacked shunts initially made - I think - contact with the block, needing me to insulate the tops against a potential short and generally it took me 3 attempts to get the block on satisfactorily and I'm scared to take it back off except to change it to a better block (even a good loop is only giving me 35-40C or so).
I doubt any of this is particularly helpful because it's as much alchemy as overclocking. I've never had a board that has such odd power balancing and although I love the peformance generally, I can't pretend to properly understand it. If I can give you anything, it's my belief that there's some kind of way to trick this board using the 'IDLE' power mode and if I ever figure it out I'll certainly post it up.
Anyway, sorry for the delay and the long essay - I realise that 'adventures with a pcb' makes for very specific bedtime reading
damn man, how many bottles of red wine did you have tonight?
so in essence, you did the following?
Associate
- Joined
- 7 Apr 2006
- Posts
- 292
- Location
- USA
Actually this is quite interesting. Thank you!
I've seen this "idle" @ 1.10v thing also. But I've only seen it happen during certain temperature ranges, and with the shape of the V/F curve a certain way.
It's easy to see this when you're NOT running a game, and have the card running at 3D clocks (e.g. by locking 1.10v on the v/f curve yourself).
Normally, here is the behavior of the card at idle, but at full 3D clocks.
At 0% voltage slider (this slider doesn't increase voltage, it just unlocks 1 higher point on the V/F curve, but also increases core clock by 15 mhz since that's part of the curve):
1.056v-1.069v should result in VREL only (boost clocks are being limited by voltage reliability), if the V/F point selected by the card is not at the end of the current tier
1.075v-1.081v should (almost always) result in VREL + vOP (maximum operating voltage has been reached).
If the temperature remains identical, the curve "should not" change shape and change the voltage points, but there's more to the curve than just temperature downclocks!
The voltage point the card uses is "often" the left most point of the current v/f tier, especially if the slider is at 100% voltage. If the current tier is 1.056v-1.075v, and it can't boost to the next (1.081v) tier, the voltage will be 1.056v (and will show VREL).
Doing a really quick test.
at full idle, but at full 3D clocks, the voltage point can be set to the RIGHT MOST point on the current tier, depending on temperature. For example, 1.056v-1.075v, it can select 1.075v, but will show VREL+VOP.
If it shows 1.075v and vrel+vop, for some reason it can't boost to 1.081v. You may be able to see this if you run a quick load test then quickly exit the program with the Afterburner graph showing.
Then as the card cools even more, the voltage jumps to the next tier, 1.081v and +15 mhz, and remains there, with the curve not changing.
But if you put a load on the card, it immediately drops to 1.056v and shows VREL.
Basically, any point on the graph on the tier BELOW the 1.10v tier, that is NOT at the end of the current tier, will show VREL.
Example:
1.056v-1.075v --->1.081v-1.10v (two tiers):
Voltage=1.056v-1.069v = VREL
1.075v-1.081v= vREL + vOP.
At 100% TDP slider, this usually causes the v/f selection to jump a tier, causing a cycle from 1.069v to 1.10v, with the curve getting shifted around.
At 100% TDP slider, I have NEVER seen vREL by itself. Only vREL + vOP, and "idle".
I have also seen the card, when it shows "idle", suddenly seem to want to use more power or heat than normal.
This only seems to happen at 1.10v. And apparently only at a certain temperature and if the curve is shaped a certain way.
As far as which rail is causing "TDP Normalized %" to trigger a power limit: I still can't be sure. it seems someway related to GPU Core NVVDD Output Power (sum), (which is the sum of the bottom GPU Core NVVDD Output Power + SRAM output power), but SRAM output power is linked to SRAM input power. And I've seen a throttle at both 100W SRAM input power, and I've seen SRAM input power get much higher than that without getting close to "TDP Normalized 108%".
It's almost like something is limiting the very last "Output power" value, but it is completely unclear where it is getting this value from. Because these output values seem to ignore all shunt mods.
Anyway, someone is currently trying an experiment with shunting PCIE Slot Power and SRC (both of them) at lower resistance than all the other shunts, to see if SRC is having an effect on this.
*edit* here is a really old Super position 4k extreme shaders run.
If you notice, super thick Perf powercap. However NONE of the primary input rails are at any rail limits whatsoever (I've exceeded some of the limits shown here just running Heaven!).
The only rails which are suspicious are SRAM Input Power and that NVVDD Output Power (sum) at the bottom, which are reporting extremely high. Now normally I would say "hey, it's SRAM input power causing the throttle!" Not so fast. I've encountered the same throttle with SRAM input power only at 97W, in Overwatch at 4k, but it was only a "blip" (momentary) throttle. And looping Heaven causes SRAM input power to exceed 112W, with only 87% normalized TDP, so it really makes you wonder what's going on. The only thing left is NVVDD Output Power (the bottom one), the second part of the sum. And I have no idea where it's getting that value from (again, this rail does NOT respond to shunt mods--it just goes up and up...).
Last edited:
I sent the card back to the place of purchase as faulty, hadn't heard anything from them for a few days and assumed they were just busy.
This morning I get a message from DPD saying I am due a delivery today. I hadn't ordered anything recently, so wasn't sure what to find in the box. But had a feeling it was what I was expecting, but actually it wasn't what I was expecting at all!
I almost cried when I opened the box!!
Here's what I found inside.
I honestly never expected to get a replacement card. I haven't actually tested the card yet. But sure it will be fine.
The burning question is, do I commence with putting the EK water block on this new card?!
Me and the girlfriend have made up, so at least there wont be any distractions!
This morning I get a message from DPD saying I am due a delivery today. I hadn't ordered anything recently, so wasn't sure what to find in the box. But had a feeling it was what I was expecting, but actually it wasn't what I was expecting at all!
I almost cried when I opened the box!!
Here's what I found inside.
I honestly never expected to get a replacement card. I haven't actually tested the card yet. But sure it will be fine.
The burning question is, do I commence with putting the EK water block on this new card?!
Me and the girlfriend have made up, so at least there wont be any distractions!
The burning question is, do I commence with putting the EK water block on this new card?!
Me and the girlfriend have made up, so at least there wont be any distractions!
No, there's a reason you were this lucky to get another card...
I sent the card back to the place of purchase as faulty, hadn't heard anything from them for a few days and assumed they were just busy.
This morning I get a message from DPD saying I am due a delivery today. I hadn't ordered anything recently, so wasn't sure what to find in the box. But had a feeling it was what I was expecting, but actually it wasn't what I was expecting at all!
I almost cried when I opened the box!!
Here's what I found inside.
I honestly never expected to get a replacement card. I haven't actually tested the card yet. But sure it will be fine.
The burning question is, do I commence with putting the EK water block on this new card?!
Me and the girlfriend have made up, so at least there wont be any distractions!
Make sure you buy a lottery ticket, you jammy bar steward!
THIS!! Leave the Air Cooler on it and be happy you dodged a bullet.
Associate
- Joined
- 19 Jan 2007
- Posts
- 2,019
- Location
- Up a Tree
Agreed, don't mess with it, you've been incredibly lucky.
You have accidental coverage, but this wouldnt be classed as accidental.