Words fail me in dealing with these damn things. Here follows a choice selection of merely the most perplexing and baffling of issues...
1) You may never again see your screen light up after a RAM or CPU upgrade.
2) RAM errors can be CPU-dependent; simply leaving you screenless on one chip, but present a blinking led on another. Now and then.
3) Sometimes your system will start and shut down, a few times... because you added a PCIE card... other times it won't care all RAM and CPU's have been upgraded.
4) When you connect the power you MUST disconnect lest it belches at you with the intrusion alarm, it will power on, spin up drives... and then shut down, after all. It merely completes those steps during which hardware failures are most likely, and quits. Each time. Well - almost
Two times (out of about 11 or so) it did NOT spring to life upon inserting the power cable. Why? Dude... if you need to ask, this PC ain't for you my friend!
5) You will not know there is a problem, what it might be, and why. Like now.... NumLock is on (and I can indeed switch it on and off), the machine responds to the power button right away (it is, I daresay, not crashed!) but there is no display on the Quadro M4000.
6) Random guesswork to just "try something" is not an option - it's the only thing left to do aside of throwing the damn thing (or yourself) out of the window.
7) I happen to have both; Now - the 7820... I bought as presumed faulty, so I can cut it some slack... well - until I replaced its motherboard. Behavior became somewhat more predictable... but still bonkers. Add a second CPU? Nope - this will fail for reasons unknowable.
8) The 7920 has been sitting here all this time from when I started to write this post. It is busily blinking what I think is the HDD led, and for all intents and purposes it looks like it is in Windows just without display; But when I hit the power button - 4 secs and it shuts down. This itself is a random response because the previous time it shut down instantly. No, it's not sleeping, the led is not "breathing"...not this time, for some reason unknown.
9) Whoever implemented DisplayPort really needs to explain how even by version 1.4a the cable still must be connected during power on, or else you can forget getting any video/signal until the next restart. It is the single most enraging regression imaginable. Not to mention guessing which port is purportedly the first one. No, nVidia has NO "rule". M4000 it's the bottom one. but on other cards it can be at the top. Or the one on the left-hand slot... this time!
10) Intrusion alarm. I switched the sound off, "Silent Mode" or somesuch... or so I was led to believe... apparently not - it still switches on, even (briefly) with the power unplugged... The T7920 is filled to the brim with thermometers (2 for each drive cage, and more) - but actually using them to detect if there's a thermal issue - no, that... that's too difficult. C'mon now, that would make SENSE! Preposterous reasonable nonsense! These are not AMD Durons with fragile crumbling cores which, when deprived of a heatsink, shall turn to ash in seconds; I think they can handle the case being open for a moment or two. The SuperMicro equivalent on which I have no choice but to continue working can deal with it... why not this overpriced Dell?
11) Apparently it's of great importance whether or not Secure Boot is enabled, as this can cause the machine to "decide" some security violation has come to pass. Like using a non-Dell video card. The horror! Indeed it is so horrible we shall not tell anyone - let 'em guess on Reddit and see what happens. That'll teach them to try and save $2000 and deny us our RRP sucker premium! Just as well, the machine can't.. you know - figure out that, heyyy I suspect the user might not be entirely satisfied with the screen not coming back on - maybe we should offer an emergency mode in which we temporarily switch it off to alert the customer that therein the problem lies...?
12) I may not, and this is true, use DIMMs of different sizes. Such a configuration is not officially supported. 1x16 and 2x32 you say??! Heresy! Burn at the stake ye will! You shall order your Official Dell Memory Modules for $499 per MB from US, oh peasant, and don't you even think of keeping your existing modules, insolent cheapskate!
But most of all - who in their right mind, when designing the... what... 8th? 10th? generation of their FLAGSHIP desktop Product for professionals... would go with "Yeah, let's make it... go dark now and then for reasons we will not explain. The error, whatever it might happen to be, shall not be logged, either. We shall thus make sure the dumb sucker - I mean, the customer has a feeling of dread, insecurity and stressful ignorance... make him have thoughts like - hmm, well, that LGA3647 socket IS very fragile, maybe I... bent a pin...? Perhaps I should reseat (and again run the risk of damaging) this ultra-fidgety CPU...?"
"What if I put the video card in a different slot?"
THE VERY FACT THIS IS EVEN A STEP TO BE CONTEMPLATED is a STRONG indicator *someone* designed things VERY badly. This is not the era of 486 with VESA local bus cards which actually require exact timings lest they literally can't function.... This was 2019-2020, machines so advanced they'd make a T1000 blush... and THIS is whence we have regressed?? And this is not (at the time) cutting edge PCIE 4... no no... even the 2nd Generation Intel scalable was still all PCIE 3. Tried and tested, cut and dried. And yet... the web's abrim with prayers to Randomitor, Overlord of Happenstance, hoping that [the other slot] will appease its capricious dictates.
And finally... Why DID Intel make PIN1 so hard to readily identify? Like, are you telling me THIS was the ONLY feasible socket design....? Not a SINGLE pin could have been arranged differently... to make the orientation readily self-evident? They got half a dozen deviations from the pin pattern on the socket... but replicate them on the CPU...? - no way! That would, like, totally take away the mystery, dude!
Anyway - I am going back to it. I am honestly believing it might have helped that it got to... "rest a little now"... maybe it will be in a better mood, who knows.
1) You may never again see your screen light up after a RAM or CPU upgrade.
2) RAM errors can be CPU-dependent; simply leaving you screenless on one chip, but present a blinking led on another. Now and then.
3) Sometimes your system will start and shut down, a few times... because you added a PCIE card... other times it won't care all RAM and CPU's have been upgraded.
4) When you connect the power you MUST disconnect lest it belches at you with the intrusion alarm, it will power on, spin up drives... and then shut down, after all. It merely completes those steps during which hardware failures are most likely, and quits. Each time. Well - almost
Two times (out of about 11 or so) it did NOT spring to life upon inserting the power cable. Why? Dude... if you need to ask, this PC ain't for you my friend!5) You will not know there is a problem, what it might be, and why. Like now.... NumLock is on (and I can indeed switch it on and off), the machine responds to the power button right away (it is, I daresay, not crashed!) but there is no display on the Quadro M4000.
6) Random guesswork to just "try something" is not an option - it's the only thing left to do aside of throwing the damn thing (or yourself) out of the window.
7) I happen to have both; Now - the 7820... I bought as presumed faulty, so I can cut it some slack... well - until I replaced its motherboard. Behavior became somewhat more predictable... but still bonkers. Add a second CPU? Nope - this will fail for reasons unknowable.
8) The 7920 has been sitting here all this time from when I started to write this post. It is busily blinking what I think is the HDD led, and for all intents and purposes it looks like it is in Windows just without display; But when I hit the power button - 4 secs and it shuts down. This itself is a random response because the previous time it shut down instantly. No, it's not sleeping, the led is not "breathing"...not this time, for some reason unknown.
9) Whoever implemented DisplayPort really needs to explain how even by version 1.4a the cable still must be connected during power on, or else you can forget getting any video/signal until the next restart. It is the single most enraging regression imaginable. Not to mention guessing which port is purportedly the first one. No, nVidia has NO "rule". M4000 it's the bottom one. but on other cards it can be at the top. Or the one on the left-hand slot... this time!
10) Intrusion alarm. I switched the sound off, "Silent Mode" or somesuch... or so I was led to believe... apparently not - it still switches on, even (briefly) with the power unplugged... The T7920 is filled to the brim with thermometers (2 for each drive cage, and more) - but actually using them to detect if there's a thermal issue - no, that... that's too difficult. C'mon now, that would make SENSE! Preposterous reasonable nonsense! These are not AMD Durons with fragile crumbling cores which, when deprived of a heatsink, shall turn to ash in seconds; I think they can handle the case being open for a moment or two. The SuperMicro equivalent on which I have no choice but to continue working can deal with it... why not this overpriced Dell?
11) Apparently it's of great importance whether or not Secure Boot is enabled, as this can cause the machine to "decide" some security violation has come to pass. Like using a non-Dell video card. The horror! Indeed it is so horrible we shall not tell anyone - let 'em guess on Reddit and see what happens. That'll teach them to try and save $2000 and deny us our RRP sucker premium! Just as well, the machine can't.. you know - figure out that, heyyy I suspect the user might not be entirely satisfied with the screen not coming back on - maybe we should offer an emergency mode in which we temporarily switch it off to alert the customer that therein the problem lies...?
12) I may not, and this is true, use DIMMs of different sizes. Such a configuration is not officially supported. 1x16 and 2x32 you say??! Heresy! Burn at the stake ye will! You shall order your Official Dell Memory Modules for $499 per MB from US, oh peasant, and don't you even think of keeping your existing modules, insolent cheapskate!
But most of all - who in their right mind, when designing the... what... 8th? 10th? generation of their FLAGSHIP desktop Product for professionals... would go with "Yeah, let's make it... go dark now and then for reasons we will not explain. The error, whatever it might happen to be, shall not be logged, either. We shall thus make sure the dumb sucker - I mean, the customer
has a feeling of dread, insecurity and stressful ignorance... make him have thoughts like - hmm, well, that LGA3647 socket IS very fragile, maybe I... bent a pin...? Perhaps I should reseat (and again run the risk of damaging) this ultra-fidgety CPU...?""What if I put the video card in a different slot?"
THE VERY FACT THIS IS EVEN A STEP TO BE CONTEMPLATED is a STRONG indicator *someone* designed things VERY badly. This is not the era of 486 with VESA local bus cards which actually require exact timings lest they literally can't function.... This was 2019-2020, machines so advanced they'd make a T1000 blush... and THIS is whence we have regressed?? And this is not (at the time) cutting edge PCIE 4... no no... even the 2nd Generation Intel scalable was still all PCIE 3. Tried and tested, cut and dried. And yet... the web's abrim with prayers to Randomitor, Overlord of Happenstance, hoping that [the other slot] will appease its capricious dictates.
And finally... Why DID Intel make PIN1 so hard to readily identify? Like, are you telling me THIS was the ONLY feasible socket design....? Not a SINGLE pin could have been arranged differently... to make the orientation readily self-evident? They got half a dozen deviations from the pin pattern on the socket... but replicate them on the CPU...? - no way! That would, like, totally take away the mystery, dude!
Anyway - I am going back to it. I am honestly believing it might have helped that it got to... "rest a little now"... maybe it will be in a better mood, who knows.
Last edited:
