Dietmar

@Svyatpro I can confirm this black screen via USB install of XP SP3. I use Firadisk for this, which works mostly. Here not. On some other notebooks I have seen the same behavior. It is the fault of USB implementation on this chipset Haswell. Install of XP SP3 on this nice notebook works only(!) via real burned XP SP3 CD in its DVD slot. Then, I get everything to work under XP SP3 on this notebook from Dell M4800. Waaooh, it is the fastest notebook, that I have ever seen fro XP Dietmar

@reboot12 Hi, "However, it is possible to boot in UEFI32" does this mean, that now we can boot XP SP3 with graphic on a compi, that has only 64 bit UEFI (no CSM at all) Dietmar

@Svyatpro The problem can be the USB. I make the same experiance of hanging at "Windows is starting" before. Make a try with a real CD-Rom, because this notebook Dell precision m4800 has an DVD drive. Use Ramseys XP SP3 from 29 November 2024 with the settings 5, E, F, M, R good luck Dietmar PS: In Bios you can switch between the 2 graphic cards. This can also be a reason for hanging.

@Svyatpro May be, that full XP SP3 can be installed on this Dell precision m4800 notebook. There are different versions of it. Please tell me all about its hardware, cpu, graphiccard, SSD or HD Dietmar

@reboot12 For XP SP3 bit32 you can also use https://msfn.org/board/topic/179059-acpi-patcher-at-boot-time/#comment-1165701 https://github.com/MovAX0xDEAD/ACPI-Patcher for to use a modded DSDT during boottime. No modd in Bios itself! This is a crazy effective way, for to start any compi with original acpi.sys. Until now, I do not see a single compi, where this methode does not work. But it is only for XP SP3 bit32 until now. Would be a nice project, to enable this also for XP SP2 bit64. We have to ask @Mov AX, 0xDEAD for help for this Dietmar

@reboot12 Yes. This is a crazy behavior of the underlying DSDT table. In summer 2019 we check everything for XP SP3 about USB3 on modern compi. And we can see, that the nearly only reason, why there was no USB3 on modern compis, is a crazy sleeping order in DSDT. All manufacturers try to put XP out of the market and so they come to this "genius" idea. This comes together with acpi2.0 version. But it does not depend on acpi.sys itself. But for to overcome this sleeping order, you have to use the modded acpi.sys or to modd the DSDT in Bios for to get a working USB3 Dietmar

@reboot12 No. The original acpi.sys works with USB3. The modded acpi.sys is only needed, because most compis with USB3 have acpi tables >1.b, which means unknown opcode for the original acpi.sys from XP SP3 Dietmar PS: I just test on the Lenovo Q180 above. This compi has USB3 and it works with any USB3 driver (Renesas, AMD 145 from Daniel, USB3 from @Mov AX, 0xDEAD ) and the original acpi.sys from XP SP3.

@casey If the modded acpi.sys is better than the original acpi.sys from XP SP3 depends on for what you use it. When you have a notebook, I think in any case the modded one is the better choice. When you run win10 on your compi, you may see some more devices in Device Manager, because of the OS hack in this modded acpi.sys. In most cases for a desktop compi before 2015, you will see no difference at all. I just test this on a Lenovo Q180 from 2012 with D2550 cpu and AMD Radeon 6450 , same result Dietmar

@casey A Bios Update or Downgrade may help. But any operation with Bios is a risk Dietmar

@casey Hi, disable everything in Bios, that you dont need for boot. All USB, sound, network, internal graphik, use graphik card, etc. Dietmar PS: I did it for myself on this crazy q1900 board. Step by step I also disabled everything in Device Manager of XP SP3. "Now I have also sound on full ACPI Multiprocessor XP SP3 on the Q1900m board. For this I have to disable "Microsoft UAA-Bustreiber für High Definition Audio" in Device Manager and after this to enable "Microsoft UAA-Bustreiber für High Definition Audio" again, crazy."

@casey At the moment, this upload does not work for me. But this acpi.sys is now also in the XP from Ramsey, good luck Dietmar

@Damnation Oh nice, how did you achieve this? Do you change something in the Source Tables from XP SP1 via @Mov AX, 0xDEAD ? I changed soso much in my last acpi.sys, I cant remember all. When you get a special Bsod, I can look for this Dietmar

Waoooh, big step forward. Just now no stopping video at youtube any longer with this version 68.14.5b of MyPal for XP Dietmar

@Outbreaker This is my last working acpi.sys for XP SP3, bit32. https://ufile.io/aijaqmds I modded such a lot of other acpi.sys, but they are always very special for ONE Bsod. This acpi.sys above will work for nearly all desktop pcs from 1999 to 2025 for XP. But for newer notebooks (starting from 2015) there are a lot of different hurdles, mostly in the underlying DSDT for its graphic card Dietmar

@N100dc Here is the link. There is a new Beta-Bios, but I have not tested. This here works for CSM and XP Dietmar https://ufile.io/zk9i1a2j

This crazy program learns via Brute Force in an ultrashort time ALL 16 tables, that are possible for an input 0 or 1 into 2 different Neurons, means AND, OR, XOR etc, just all of them Dietmar import numpy as np from itertools import product # To generate all binary combinations and weight combinations # Initialisierung der Schwellenwerte lower_threshold = 0.8 upper_threshold = 1.2 # Trainingsdaten (Inputs für das XOR-Problem und andere) inputs = [[0, 0], [0, 1], [1, 0], [1, 1]] # Alle möglichen Zieltabellen (16 Kombinationen) all_possible_targets = list(product([0, 1], repeat=4)) # Mögliche Werte für die Gewichte (in Schritten von 0.1) weight_values = np.arange(-1.0, 1.1, 0.1) # Trainingsloop für jede mögliche Zieltabelle for table_index, targets in enumerate(all_possible_targets, start=1): print(f"\n=== Wahrheitstabelle {table_index}: Targets = {targets} ===") # Initialisieren der Startwerte bias_list = [0.0, 0.7, 0.9] # Bias nur mit den Werten 0.0, 0.7, 0.9 network_trained = False final_weights = None # Iterate over bias values for bias in bias_list: print(f"Versuch mit Bias {bias}:") # Teste alle Kombinationen der Gewichte for weight_combination in product(weight_values, repeat=2): current_weights = np.array(weight_combination) all_correct = True for input_vector, target in zip(inputs, targets): # Berechnung der gewichteten Summe inkl. Bias weighted_sum = np.dot(input_vector, current_weights) + bias # Aktivierungsfunktion (Schwellenwertfunktion mit zwei Schwellenwerten) output = 1 if lower_threshold < weighted_sum < upper_threshold else 0 # Überprüfe, ob die Ausgabe korrekt ist if target != output: all_correct = False break # Kein Erfolg mit diesen Gewichten; abbrechen # Wenn alle Ausgaben korrekt sind, speichere die Gewichte und Bias if all_correct: network_trained = True final_weights = current_weights break if network_trained: print(f"Das Netzwerk hat Wahrheitstabelle {table_index} erfolgreich gelernt.") print(f"Gewählte Gewichte: {final_weights}") print(f"Gewählter Bias: {bias}") break # Weiter zur nächsten Wahrheitstabelle if not network_trained: print(f"Das Netzwerk hat Wahrheitstabelle {table_index} nicht gelernt.") === Wahrheitstabelle 1: Targets = (0, 0, 0, 0) === Versuch mit Bias 0.0: Das Netzwerk hat Wahrheitstabelle 1 erfolgreich gelernt. Gew&auml;hlte Gewichte: [-1. -1.] Gew&auml;hlter Bias: 0.0 === Wahrheitstabelle 2: Targets = (0, 0, 0, 1) === Versuch mit Bias 0.0: Das Netzwerk hat Wahrheitstabelle 2 erfolgreich gelernt. Gew&auml;hlte Gewichte: [0.1 0.8] Gew&auml;hlter Bias: 0.0 === Wahrheitstabelle 3: Targets = (0, 0, 1, 0) === Versuch mit Bias 0.0: Das Netzwerk hat Wahrheitstabelle 3 erfolgreich gelernt. Gew&auml;hlte Gewichte: [ 0.9 -1. ] Gew&auml;hlter Bias: 0.0 === Wahrheitstabelle 4: Targets = (0, 0, 1, 1) === Versuch mit Bias 0.0: Das Netzwerk hat Wahrheitstabelle 4 erfolgreich gelernt. Gew&auml;hlte Gewichte: [ 9.00000000e-01 -2.22044605e-16] Gew&auml;hlter Bias: 0.0 === Wahrheitstabelle 5: Targets = (0, 1, 0, 0) === Versuch mit Bias 0.0: Das Netzwerk hat Wahrheitstabelle 5 erfolgreich gelernt. Gew&auml;hlte Gewichte: [-1. 0.9] Gew&auml;hlter Bias: 0.0 === Wahrheitstabelle 6: Targets = (0, 1, 0, 1) === Versuch mit Bias 0.0: Das Netzwerk hat Wahrheitstabelle 6 erfolgreich gelernt. Gew&auml;hlte Gewichte: [-0.1 1. ] Gew&auml;hlter Bias: 0.0 === Wahrheitstabelle 7: Targets = (0, 1, 1, 0) === Versuch mit Bias 0.0: Das Netzwerk hat Wahrheitstabelle 7 erfolgreich gelernt. Gew&auml;hlte Gewichte: [0.9 0.9] Gew&auml;hlter Bias: 0.0 === Wahrheitstabelle 8: Targets = (0, 1, 1, 1) === Versuch mit Bias 0.0: Versuch mit Bias 0.7: Das Netzwerk hat Wahrheitstabelle 8 erfolgreich gelernt. Gew&auml;hlte Gewichte: [0.2 0.2] Gew&auml;hlter Bias: 0.7 === Wahrheitstabelle 9: Targets = (1, 0, 0, 0) === Versuch mit Bias 0.0: Versuch mit Bias 0.7: Versuch mit Bias 0.9: Das Netzwerk hat Wahrheitstabelle 9 erfolgreich gelernt. Gew&auml;hlte Gewichte: [-1. -1.] Gew&auml;hlter Bias: 0.9 === Wahrheitstabelle 10: Targets = (1, 0, 0, 1) === Versuch mit Bias 0.0: Versuch mit Bias 0.7: Versuch mit Bias 0.9: Das Netzwerk hat Wahrheitstabelle 10 erfolgreich gelernt. Gew&auml;hlte Gewichte: [-1. 1.] Gew&auml;hlter Bias: 0.9 === Wahrheitstabelle 11: Targets = (1, 0, 1, 0) === Versuch mit Bias 0.0: Versuch mit Bias 0.7: Versuch mit Bias 0.9: Das Netzwerk hat Wahrheitstabelle 11 erfolgreich gelernt. Gew&auml;hlte Gewichte: [-2.22044605e-16 -1.00000000e+00] Gew&auml;hlter Bias: 0.9 === Wahrheitstabelle 12: Targets = (1, 0, 1, 1) === Versuch mit Bias 0.0: Versuch mit Bias 0.7: Versuch mit Bias 0.9: Das Netzwerk hat Wahrheitstabelle 12 erfolgreich gelernt. Gew&auml;hlte Gewichte: [ 0.1 -0.1] Gew&auml;hlter Bias: 0.9 === Wahrheitstabelle 13: Targets = (1, 1, 0, 0) === Versuch mit Bias 0.0: Versuch mit Bias 0.7: Versuch mit Bias 0.9: Das Netzwerk hat Wahrheitstabelle 13 erfolgreich gelernt. Gew&auml;hlte Gewichte: [-1.00000000e+00 -2.22044605e-16] Gew&auml;hlter Bias: 0.9 === Wahrheitstabelle 14: Targets = (1, 1, 0, 1) === Versuch mit Bias 0.0: Versuch mit Bias 0.7: Versuch mit Bias 0.9: Das Netzwerk hat Wahrheitstabelle 14 erfolgreich gelernt. Gew&auml;hlte Gewichte: [-0.3 0.3] Gew&auml;hlter Bias: 0.9 === Wahrheitstabelle 15: Targets = (1, 1, 1, 0) === Versuch mit Bias 0.0: Versuch mit Bias 0.7: Versuch mit Bias 0.9: Das Netzwerk hat Wahrheitstabelle 15 erfolgreich gelernt. Gew&auml;hlte Gewichte: [0.1 0.3] Gew&auml;hlter Bias: 0.9 === Wahrheitstabelle 16: Targets = (1, 1, 1, 1) === Versuch mit Bias 0.0: Versuch mit Bias 0.7: Versuch mit Bias 0.9: Das Netzwerk hat Wahrheitstabelle 16 erfolgreich gelernt. Gew&auml;hlte Gewichte: [-2.22044605e-16 -2.22044605e-16] Gew&auml;hlter Bias: 0.9

@MilkChan Try this one, good luck Dietmar https://ufile.io/vknb4j2p

@Outbreaker I build some special acpi.sys for crazy notebooks. But most they work only on one or 2 different laptops. So, until now the acpi.sys that you have is still the best for Desktop compis Dietmar

@aliktz This files are all for XP SP2 bit64. You tell, that you installed XP x86 32bit. Make a try with a normal XP SP3 install, choosing Ramseys XP Dietmar

@aliktz Here is new link. I update also acpi.sys bit64 to its very last version Dietmar https://ufile.io/vknb4j2p

Until now, I have no idea of Python. But with the help of ChatGPT I succeed to make my Ki learn XOR. For this, I use 3 Neurons with 2 thresholds. Together with the Hebb learning (1949(!)), 2 neurons that fire together, wire together. And if they fire randomally to each other, the weights are lowered. I succeed via Hebb with XOR Dietmar PS: You can copy and paste this code into ideone.com choose Python 3 and hit run. import numpy as np # Initialisierung der Schwellenwerte lower_threshold = 0.8 upper_threshold = 1.2 # Lernrate learning_rate = 0.1 # Trainingsdaten (XOR-Problem) inputs = [[0, 0], [0, 1], [1, 0], [1, 1]] targets = [0, 1, 1, 0] # Trainingsloop mit max. 1000 Iterationen max_iterations = 1000 epoch = 0 network_trained = False start_weights = None final_weights = None all_epoch_outputs = [] # Store outputs of all epochs for debugging and transparency while epoch < max_iterations: epoch += 1 all_correct = True # Flag, um zu überprüfen, ob alle Ausgaben korrekt sind current_weights = np.random.rand(2) # Zufällige Startgewichte if epoch == 1: # Die erste Iteration nach Initialisierung start_weights = current_weights # Speichere die Startgewichte epoch_outputs = [] # To store outputs of this epoch for input_vector, target in zip(inputs, targets): # Berechnung der gewichteten Summe weighted_sum = np.dot(input_vector, current_weights) # Aktivierungsfunktion (einfache Schwellenwertfunktion) output = 1 if lower_threshold < weighted_sum < upper_threshold else 0 # Fehlerberechnung error = target - output # Wenn ein Fehler vorliegt, dann weise die Gewichte an if error != 0: all_correct = False current_weights += learning_rate * error * np.array(input_vector) epoch_outputs.append((input_vector, output, target)) # Save each iteration's output all_epoch_outputs.append(epoch_outputs) # Überprüfe, ob alle Ausgaben korrekt sind if all_correct: network_trained = True final_weights = current_weights # Speichere die finalen Gewichte break # Stoppe, wenn alle Ausgaben korrekt sind # Wenn XOR nach 100 Iterationen nicht gelernt wurde, setze neue zufällige Startgewichte if epoch % 100 == 0: # 100 statt 20 print(f"Nicht funktionierende Startgewichte: {start_weights}") start_weights = np.random.rand(2) # Setze neue Startgewichte if network_trained: print(f"Das Netzwerk hat XOR korrekt nach {epoch} Iterationen gelernt.") print(f"Die Working Startgewichte waren: {start_weights}") print(f"Die finalen Gewichte sind: {final_weights}") else: print(f"Das Netzwerk hat XOR nach {epoch} Iterationen nicht korrekt gelernt.") # Testen des Netzwerks nach den Lern-Iterationen print("\nFinal Test Output:") for input_vector, target in zip(inputs, targets): weighted_sum = np.dot(input_vector, final_weights) output = 1 if lower_threshold < weighted_sum < upper_threshold else 0 print(f"Input: {input_vector}, Target: {target}, Output: {output}") # Optionally, print out the outputs of each epoch for transparency print("\nEpoch Outputs:") for epoch_index, epoch_outputs in enumerate(all_epoch_outputs): print(f"Epoch {epoch_index + 1}:") for input_vector, output, target in epoch_outputs: print(f" Input: {input_vector}, Output: {output}, Target: {target}")

@reboot12 Here is the wished acpi.sys bit64 for XP SP2. I just do the hack for C0140010 ==> 00000000 at 3 places, following the idea from skull. Dont know, how serious this Bsod is, anyway now it should be gone Dietmar https://ufile.io/hw7xnwtv

@reboot12 The important part of this Bsod is C0140010. It means, that via acpi an invalid region for the target was specified. C0140010, STATUS_ACPI_INVALID_REGION I never saw this Bsod before in XP SP3, bit32. In the evening I am at home and try to make a hack for it Dietmar

@MilkChan Hi, try this one for XP SP3. I have others with some special modd for notebooks, but this should work for you. Good luck Dietmar https://ufile.io/aijaqmds

Hi, may be the problem is the strange behavior on some nvme devices, when you install there XP SP3. So, make a full install on a normal harddisk for this notebook. If there is no harddisk, you can make the XP install to a connected USB3 drive (harddisk). For this, enable the life USB on Ramsey setup. Later you can copy your whole XP install back to your nvme device. Dont forget to install, before USB install, the nvme driver. Good luck Dietmar