wutske

Decided to keep the VRMs like they are but with loads of vaseline on them. Going to order some dry ice now and tomorrow I'm going to have some fun

Did some preparations today and a test fit, it's a tight fit, that one cap is going to get very cold. Insulation won't be easy with the angled side. Should I also cover the VRM's or not ? Seems like a bad idea since they don't have a heatsink so all heat will be trapped under the insulation.

1) Awesome news ? . Glad to know that it won't harm my submissions. 2) Can this be added to the FAQ ? Not sure if many people still use Windows 7, but it's nice to know since the error seems kind of fatal (and was pretty much the reason I gave up on using it a while ago).

It's actually the first time I have Benchmate working on W7 so not sure if everything is normal or not. 1) I'm getting a "Could not verify CPUID brand string" warning in my results 2) When I launched BM right after starting Windows, it won't start and I get an error about a bad ACPI implementation, however, if I then wait a couple of minutes (it varies a lot), then BM will just launch without any issues. I assume it's some sluggishness in Windows right after the startup, but maybe it's worth mentioning in the error message that it might be worth to retry after a couple of minutes. Some logs related to this issue LOG START at 2023-10-29 17:46:57 ---------------------- BenchMate 11.3.0 DPI Awareness used: System Init Status: VERIFYING INSTALLATION Init Status: CLEANING UP Removing old process log files. 2 old process log files successfully deleted. Init Status: PREPARING CLIENT Preparing client. Init Status: INTIALIZING DRIVER Initializing timer. Loading driver file: C:\Tools\bm-manual-11.3.0\bin\bmrbt64.sys Installing the driver (Timeout: 10000 ms) Creating driver service Driver status: Stopped Starting driver service Could not start driver service: If an MM error is returned which is not defined in the standard FsRtl filter, it is converted to one of the following errors which is guaranteed to be in the filter. In this case information is lost, however, the filter correctly handles the exception. [#557] Driver installation failed: If an MM error is returned which is not defined in the standard FsRtl filter, it is converted to one of the following errors which is guaranteed to be in the filter. In this case information is lost, however, the filter correctly handles the exception. [#557] Driver initialization failed! Init Error: Driver installation failed! The ACPI PM Timer could not be initialized due to an incompatible implementation with your system's ACPI specification. Please consider filing a bug report to make us aware of that problem. CPUID information: - Vendor: GenuineIntel - Brand: Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz - Family 0x6, Model 0x17, Stepping 7 LOG START at 2023-10-29 17:43:56 ---------------------- BenchMate 11.3.0 Initializing components: => DPI Awareness DPI Awareness used: System => QPC timestamp => Fonts => Adjusting window style => WMI => Query system data Could not fetch result field from queried row (Query: SELECT * FROM Win32_PhysicalMemory, Result field: SMBIOSMemoryType, Row: 1, Error: 0x80041002) Could not fetch result field from queried row (Query: SELECT * FROM Win32_PhysicalMemory, Result field: SMBIOSMemoryType, Row: 2, Error: 0x80041002) => cURL => Benchmark factory => Submission object => CPU affinity map => Preset UI => Single node found: 1 => Single node found: 1 => Loading client config file: client.json => Loading apps config file: apps.json => Loading result meta files => Sensor thread => Process thread => Timer Skew Test thread => Init thread Theme selected: BenchMate Init thread running: 0x0000000000000224 New initialization update status: Installing driver Initializing Reliable Benchmark Timer driver Reliable Benchmark Timer successfully initialized. HPET Mode: HPET32 HPET Frequency: 14.32 MHz Resuming timer skew test thread New initialization update status: Resuming process thread Configured system timers for skew test: New initialization update status: Initializing HWiNFO Process thread running: 0x000000000000020C => Hotkeys => HPET32, 14.32 MHz Initializing HWiNFO in sensor thread Hook extra config not found in benchmark's extra config map: 0 3 hotkeys successfully registered! => ACPI PM, 3.58 MHz Sensor thread running: 0x0000000000000200 Hook extra config not found in benchmark's extra config map: 0 Components successfully initialized. => QPC TSC, 2.93 MHz CPUID Information: Hook extra config not found in benchmark's extra config map: 0 - Vendor: GenuineIntel Hook extra config not found in benchmark's extra config map: 0 Skew Test: - Brand: Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz Hook extra config not found in benchmark's extra config map: 0 - Family 6h, Model 17h, Stepping 7 CPU Architecture detected: SMBIOS 2.4: - CPU Manufacturer: Intel - CPU PartNumber: - CPU Version: Intel(R) Core(TM)2 Quad CPU - CPU SocketDesignation: Socket 775 - CPU Core Count: 0 Hook extra config not found in benchmark's extra config map: 0 - CPU Core Enabled: 0 Hook extra config not found in benchmark's extra config map: 0 - CPU Thread Count: 0 Hook extra config not found in benchmark's extra config map: 0 - CPU Thread Enabled: 0 Hook extra config not found in benchmark's extra config map: 0 Hook extra config not found in benchmark's extra config map: 0 TSC frequency measured by driver: 3000.57087 MHz (3000570868) Loading HWiNFO library: C:\Tools\bm-manual-11.3.0\bin\HWiNFO64.dll Initializing HWiNFO. HWiNFO bus frequency detection is determined as unreliable Fetching CPUs. CPU #0: => ES: no => Cores: 4 Fetching memory modules. Fetching system sensors. #0: System: GIGABYTE EP45-DS3R #1: CPU [#0]: Intel Core 2 Quad Q9550 Postprocessing CPU main sensor: #0 - Frequency sensors ----- -- 2999.66 => Core 0 Clock (Frequency/Voltage: Core/P-core found) -- 2999.66 => Core 1 Clock (Frequency/Voltage: Core/P-core found) -- 2999.66 => Core 2 Clock (Frequency/Voltage: Core/P-core found) -- 2999.66 => Core 3 Clock (Frequency/Voltage: Core/P-core found) -- 352.90 => Bus Clock (Bus frequency found) -- 2482.52 => Core 0 Effective Clock (Effective Clock: Core/P-core found) -- 123.05 => Core 1 Effective Clock (Effective Clock: Core/P-core found) -- 549.16 => Core 2 Effective Clock (Effective Clock: Core/P-core found) -- 86.89 => Core 3 Effective Clock (Effective Clock: Core/P-core found) CPU sensor index for Ring/LLC frequency not found! - Load sensors ----- -- 84.51 => Core 0 Usage (Core/P-core found) -- 1.41 => Core 1 Usage (Core/P-core found) -- 14.08 => Core 2 Usage (Core/P-core found) -- 0.00 => Core 3 Usage (Core/P-core found) -- 84.51 => Max CPU/Thread Usage -- 25.00 => Total CPU Usage (Single sensor index found) -- 100.00 => On-Demand Clock Modulation CPU load measurement type: PerCPU (4P/0E) #2: CPU [#0]: Intel Core 2 Quad Q9550: DTS Postprocessing CPU Intel DTS sensor: #0 - Temp sensors ----- -- 40.00 => Core 0 (Core/P-core found) -- 40.00 => Core 1 (Core/P-core found) -- 43.00 => Core 2 (Core/P-core found) -- 33.00 => Core 3 (Core/P-core found) -- 60.00 => Core 0 Distance to TjMAX -- 60.00 => Core 1 Distance to TjMAX -- 57.00 => Core 2 Distance to TjMAX -- 67.00 => Core 3 Distance to TjMAX -- 43.00 => Core Max (Single sensor index found) CPU temperature measurement type: PerCPU (4P/0E) #3: CPU [#0]: Intel Core 2 Quad Q9550: C-State Residency #4: Memory Timings Postprocessing memory sensor: -- 441.13 => Memory Clock -- 1.25 => Memory Clock Ratio (x) -- 5.00 => Tcas (T) -- 5.00 => Tcas (T) -- 5.00 => Trcd (T) -- 5.00 => Trcd (T) -- 5.00 => Trp (T) -- 5.00 => Trp (T) -- 15.00 => Tras (T) -- 15.00 => Tras (T) #5: GIGABYTE EP45-DS3R (ITE IT8718F) #6: GPU [#0]: NVIDIA GeForce GTX 770: GPU #0 found: NVIDIA GeForce GTX 770 Postprocessing GPU main sensor: -- 135.00 => GPU Clock (Core frequency found) -- 162.00 => GPU Memory Clock (Memory frequency found) -- 405.00 => GPU Video Clock -- 135.00 => GPU Effective Clock #7: Windows Hardware Errors (WHEA) HWiNFO successfully initialized. Init thread exiting Using HWiNFO CPU detection: Core2 Quad Q9550 => Core 2 Quad Q9550 | Intel => Intel Client initialization successfully notified as done! Benchmark processes running: 2 Could not get file version of HWiNFO driver: C:\Users\BENCHI~1\AppData\Local\Temp\HWiNFO64A_W7_187.SYS Maximum allowed QPC skewing: 10.60 ms - HPET Skewing: 0.001676191 ms (Allowed) - QPC Skewing: -0.059580583 ms (Allowed) - QPC Drift: -0.009967582 ms/second Timer skew test thread exiting Timer skew test update success 4 hotkeys successfully unregistered! HWiNFO successfully released. Sensor thread exiting Process update canceled! No filter available to handle the IRP type: 0 Process thread exiting Exiting driver with 4 remaining processes. Driver still in use. The last process will need to remove the driver.

https://hwbot.org/submission/5141744_wutske_cpu_frequency_pentium_e2200_4015.86_mhz 1.88V on air . Somehow I even got it at 1.92V stable and cool enough for Wprime: https://hwbot.org/submission/5141760_wutske_wprime___32m_pentium_e2200_21sec_453ms Those were either outside or on the window sill during winter to get low enough ambient temps to get to that result ? I didn't run it for hours on that voltage and that CPU even survived the worst delidding ever. These things are though for short amounts of time. All CPU's are different. You can also easily kill the early Core i's when your QPI voltage was to far off the VDDR. Core2Duo's can also die easily if your GTL voltage are outside a specific range. However, these CPUs died after months of running at those settings. You could potentially kill them instantly with these settings, but the thing is, you didn't . The QPI and GTL voltage are very specific ones that either depend on other voltage or other voltage depend on them. VCore is usually pretty straightforward. The most important thing is, yours still works fine, so you didn't kill it. If you want to go into a rabbit hole, your VCore is absolutely not stable at all and it spikes and drops a lot. You can check out this video from Bildzoid : Basically, a 1.15V average Vcore is in reality a minimum voltage of 1.10V to a maximum of 1.27V and that's under a sustained load. With more varying loads it'll overshoot and undershoot a lot more. Voltage regulation is reactive: if there's load, the voltage drops (vdroop) and then the regulators allow more current to counter the voltage drop, but if the load suddenly drops again, it'll overshoot because it was allowing enough current to flow through to support a high load.

Most CPU's under normal circumstances can survive like 10 years before they stop working properly. Maybe yours will die in 9 years instead of 10 and maybe it'll overclock 50Mhz less, but it's hard to say. There's always some degradation, even when you don't overclock, overclocking just overclocks the speed of degradation. I wouldn't worry too much, your motherboard will either die before your CPU or you've already replaced the CPU by the time it would have died, CPUs just don't die that often and modern CPU's are even designed to run at 95-100°c (and Zen5's overtemp protection sometimes only kicks in at 115°c). Also, I've done 1,7V on air on a Core2Duo, thing still works fine.

The thing about taxes was a valid point, it made me lean heavily towards the ecopot ... but in the end I decided it might be worth the gamble and guess I won this time ? Almost ready to get started with dry ice. //edit: the pot arrived fully assembled and I leak tested it, hence the water and copper stains on the inside.

Aight, haven't had the time to get started, so here I am again ? I'm looking into buying a pot to do dry ice cooling. Currently have my eye on two Elmor's Volcano Bartx's ECOPot The Volcano obviously will perform better compared to the ECOPot, however, I'm not really sure if the Volcano is going to work with dry ice at all. First of all, larger pieces of dry ice won't make it down into the pot due to the many fine openings but most importantly, there doesn't seem to be any seal at all between the core and the outer tube, so all the acetone will probably just leak out (and pretty much any rubber seal will decay due to the alcohol and cold temperatures). Am I on the right track or is there some magic thing I'm missing ?

I had an MSI board which also did something similar (though I didn't exactly link it to the CPU temperature). Turned out that both BIOS chips were pretty much EOL and the board's dual BIOS setup probably had a hard time working with that situation. Though there could be multiple explanations, e.g; there's a bad contact somewhere (a slightly bent CPU pin) or just some bad solder and as soon as it heats up (and thus expands) it makes a good contact and your board boots again. Or another reason might be a bad capacitor which might take ages to charge up and supply enough voltage (though that would normally also result in further instability). Can you maybe check the VCore with a multi-meter while it's stuck ? Depending on the CPU it should be somewhere in the ~1.1V to ~1.3V range.

00 is indeed usually a dead CPU. Does it show anything before going to 00 or just immediately ? Maybe it's a good idea to check for bent pins or anything else unusual

Just out of curiosity, why would RAM latency impact your mouse input delay ? At 3600MT/s CL9, we're talking about a (RAM) latency of 5 nanoseconds. A mouse polling at 4kHz has a "latency" of 250 nanoseconds ... so even if it passes through the memory, you won't feel the difference if it's a couple of nanoseconds slower. What might make a difference is which port you're connected through. Zen4 has on-die USB support, so you want your mouse connected to one of those USB3 ports (and nothing else, no hubs or anything). Other USB ports are provided through the PCH and it has a single PCIe 4x lane to the CPU through which it also has to send stuff like wifi, m.2, ... We're still only talking about a very small delay, but I can tell from experience that running a Logitech G mouse and a USB DAC from a USB3.0 hub is horrible...

I'm looking for a simple pot for dry ice cooling, nothing too fancy, nothing too expensive either. Something like the BartX eco pot is what I'm looking for: https://bartxstore.com/shop/cpu-ln2-pot-eco/ It needs to have mounting hardware for at least S775, S115x and AM3.

I'm certainly no expert in the matter, but I've followed a lot of builds on XS eons ago, though not a lot seems to have stuck ?. Unfortunately it's also pretty much illegal to do this stuff in Belgium without the proper certifications, so I've never built on myself. Your capilary tube is now probably optimized for a 200W load, going to 260W is a 30% increase which is not insignificant. Apparently this softwarw could help you with guestimating the right length: https://www.tecumseh.com/en/eu/landing-pages/sel_soft_v4_0/ There's a lot of good threads on XS about single stages, just spend some time reading there, they're really educational from time to time.

Is that -46°c under load ? And what exactly do you mean that it can't hold load at 1.95V ? What temperatures are you getting then ? Do you have any idea about the difference in load ? Those older Intel's don't report CPU power consumption but a kill-a-watt can give you a rough idea about the extra heat they are generating. I don't know what other CPUs you are using, but single cores usually consume less, so are you sure that your system isn't suffering from a lack of load ? Do you keep an eye on the return to the compressor to make sure you don't return liquid instead of gas ? Generally, bigger is better ? , so the Embraco should perform better as long as your condenser can handle it.

Afaik only v3 was limited to 32-bit, but there is a simple workaround for that ? I never had a pro-version and it always opens a browser window with the scores, offline never worked for non-pro versions.

It's probably easier to get started with AMD Ryzen Master since you won't need to bother with MSI specificalities at first: https://www.amd.com/en/technologies/ryzen-master The built-in "AI" overclock profiles that some motherboard manufacturers add are very generalistic and are designed to work with bad silicon too, so they usually overvolt your CPU way too much, which is probably why your CPU runs warm (no worries, it can go to 90°c before it throttles and slows down again). Very simply put, your CPU's speed is defined by two factors: the base clock (BCLK) and the multiplier. Increasing the multiplier is the easiest, but also the most course way to do it. Increasing the multiplier by 0,5 on a BCLK of 100MHz means you're already running 50MHz higher than normal. The BCLK can be increased in fine steps, going from 100MHz to 100,1Mhz with a multiplier of 49 (your CPU's max boost multiplier out of the box), means you're overclocking 4,9MHz. Increasing the BCLK also increases the speed at which your RAM runs and (if not locked in the BIOS) will also increase the frequency of the PCIe bus (which will quickly result in stability issues and might also lead to data corruption since you're disk is also running on PCIe). Now, at a certain point your system will probably crash because you pushed it too hard, usually you can improve stability by slightly increasing the VCore (CPU voltage), but mind you that this doing so will increase power consumption, heat output and if pushed too high, can damage your CPU. Those are the very basics, but things are bit more complex with turbo boosting nowadays. There's a lot of information to be found on Youtube on how to use Ryzen Master, I would certainly advise to have a look there ? .

Ow, so achievements can go back down again ? No time for slack then ?

Took me a couple reads to undestand the issue ?. According to the achievement progress you have 89 golden cups but at this time you only have 83 golden cups. My best guess is that at one point you did actually have 89 golden cups but you just lost them to someone else. It makes sense that the achievement doesn't go down (otherwise, if you'll loose achievements which is kind of weird). Your links are just going to the achievements page, so it's not clear which achievements you're talking about when you're just give us a link to that page. Anyhow, if you think an achievement is not updated properly, you can just click on the achievement and hit the "recalculate achievement" button, most of the times that works (but be patient, my best guess is that you're put on a queue for processing so it takes some time).

Ah okay, well ... that explains the weird rankings and results Did all my tests single core, was great to see how hard I could push the 2500K with only a single core (and I'm not ranking that bad after all). Anyhow, since it's for testing purposes only I'll leave them as is

I'm a bit confused about this competition. Are we supposed to run the single core test (or limit the numer of threads to 1) or is this an all-core test?

Kind of want to pick this one up again just to be 100% sure ?. So it's okay if I have for example a 4-core CPU and disable 3 cores and overclock that last one to the max ? It kind of feels like cheating but having only 1 core make CPU-z validation a lot easier and it also gets a lot higher clocks too. The same question applies to single-threaded benches like SuperPi too. If it's allowed, would it be wise to put it on the rules pages ?

If you set the power limit to 85W then you're pretty much going to hit the power limit throttling all the time. PL1 and PL2 are values that define how much power your CPU can draw, so if you limit it to 85W then you're setting a much stricter limit. PL2 is usually a higher number and pretty much defines how much and how long (Tau) your CPU can throttle. So if Tau is set to 20s and 250W, then you're CPU will be consuming roughly 250W for a maximum of 20s before it's throttled back to PL1. If you have sufficient cooling capacity, setting PL2 and PL1 to very high numbers (or unlimited/unconstrained) then you won't hit the power limit anymore. This is a good article: https://www.anandtech.com/show/13544/why-intel-processors-draw-more-power-than-expected-tdp-turbo The TDP is not entirely the same as your CPU's power consumption, in fact, it's usually lower since TDP defines the amount of heat the CPU will generate, but heat is just a byproduct of work being done. Now, how does undervolting allow you to go much faster; the power consumption of your CPU is defined by two values: the VCore and the frequency at which it's running (I'm just going to ignore the GPU and memory controller and also the fact that temperatures also have a play in this). The higher the VCore, the more heat it generates. The higher the frequency at which it runs, the more heat it generates. Let's say we have a completely imaginary example, your 13900K at 1,3V running at 5,5GHz will consume 250W of power, which is the PL2 I mentioned before. Let's say we drop the VCore to 1,2V and keep the frequency at 5,5GHz, the power consumption drops to 220W (again, I make these numbers up on the spot) but your PL2 allows your CPU to consume 250W. This means your CPU can boost to higher clocks, let's say 5,7GHz, et voila, free performance ?. So basically, by dropping the VCore, you can actually have a CPU that can run faster (as long as it's stable at that VCore of course). Obviously, finding the right VCore at which you can maximize the maximum frequency is not easy and it'll take a lot of time to get to the sweet spot. //edit: about the temperatures: 90°c nowadays is kind of okay-ish. It's not great for the silicon of course when you run it like that 24/7 but a normal desktop isn't running at full load 24/7 either. Not sure if it's summer where you are, but keep in mind that in the summer your room temp might be 10°c higher so it might still throttle on hot summer days.

Apart from the high temperatures, the power limit actually means it's drawing more power than it likes to, so adding more volts will make this worse. Adding more volts also makes it run a lot hotter and 13900K are already incredibly hot and hard to cool. So I would suggest finding an overclock that works at lower voltages. With a 13900K, you can also achieve beter performance by undervolting it so that it power consumption lowers and so that it can turbo for a longer time. Not much experience with XTU, but there should probably a setting to change the turbo power limits (PL1/PL2), though at these temps, I think it's best to not touch it since your cooler already can't handle the power output.

The sad thing about this is that Futuremark has done a great work to counter the LOD tweaks in the last versions of older 3DMark titles and because of this it's pretty much impossible to beat older scores with identical hardware, but that's just how it is ?‍♂️ . With overclocking becoming more popular (and a lot of money being involved too), it makes a ton of sense that Futuremark wants to focus on delivering a score that represents a workable and playable gaming experience, setting your LOD -10 doesn't fit in that and having your view blocked by a massive plane of corrupt vectors isn't either. 2D benchmarks like SuperPi, wprime, etc ... do a lot of verifications to ensure that what comes out of your CPU and memory is actually correct because that's what counts, being the fastest to get to a specific result (and not some random numbers). For 3D benchmarks that's a lot more difficult to do though. What I'm curious about is which benchmarks have screenshot rules that are deviating, afaik for most benchmarks you need a full screen screenshot with CPU-z open (cpu & memory). Anything else is usually benchmark specific.

Windows 10 and 11 aren't constantly phoning home and the cost of that is probably minimal. It's not all about overclocking and fast hardware, you also have to optimize your OS. Windows XP is great for SuperPi but you still need to tweak it to get top scores. Same goes for other OS'es; shuting down and removing stuff you don't need during the benches can already have a positive impact on scores. Same goes for running benches at higher priorities (so that the "phoning home" doesn't impact your score that much). There's also no "golden" OS for each benchmark, it depends on the hardware you use, the drivers and even the BIOS can give different results depending on the OS.