Tech Junkie heeft een review van de Videologic DigiTheater Decoder online gezet. Dit doosje verwerkt digitale AC-3 Dolby Digital 5.1 signalen via de S/PDIF ingang, of Dolby Prologic gecodeerde audio via de normale analoge ingang. Vervolgens dient het apparaat op een versterker aangesloten te worden:

The DigiTheatre performs its job admirably, receiving a digital audio source either through S/PDIF coaxial or Toslink optic fibre, its onboard Digital Signal Processing (DSP) chip churns through and delivers very good decoding results. Dolby Prologic is received through the stereo analogue line in of the decoder and processed.

The DigiTheatre can adjusts its decoding routine according to your room's existing speaker set up. For example if you have the full blown 5.1 speaker system installed then the decoding box can be set up to take advantage of a complete 5.1 speaker system. If on the other hand, you are missing a center speaker, the decoding box can be set up to create a "phantom" center speaker by channelling the center speaker's output to the two front speakers. On the other extreme if your strapped for cash, and have a plain 2-speaker stereo set up the DigiTheatre can be set for that configuration too.

A special 9-conductor "octopus" socket is on the back of the decoder, from this the full 6 line output is delivered to an external amplifier. This rather thick octopus cable supplies the front, rear, AC-3 subwoofer, and center speaker outputs.

Op het message board van JC heeft iemand een aantal interessante database server benchmarks gepost. De K6-III met 1Mb L3 cache blijkt veruit de hoogste score te behalen (2855 records/second op 400MHz), sneller dan de Celeron 400 (2414 records/sec) en PIII-400 (2361 records/sec). JC verklaart de grote invloed van een lage L2 cache latency:

Wow, reading this really gives you an idea about how much low cache latencies are favored in the server world -- with half as much cache and on a 33% slower bus, the Celeron-400 (32K I/D L1, 3 cycle latency; 128K L2 cache, supposedly around 11 cycles in latency) managed to inch past its bigger brother, PII-400 (32K I/D L1, 3 cycle latency; 512K L2, 27 cycle latency, half the clock of the cpu). K6-3-400 (64K I/D L1, 2 cycle latency; 256K L2, 13 cycles latency according to one source, but possibly less according to one particularly bright hands-on guy; motherboard L3, generally 100MHz) was 18% beyond even the Celeron-400, probably helped along to some degree by a decently sized motherboard (L3) cache.

Actually, I should note that the motherboard cache seems to help along the K6-2 a lot here -- the K6-2-350 (64K I/D L1, 2 cycle latency; motherboard L2, generally 100MHz) is less than 10% slower than that Celeron-400, I just wish he had a K6-2-400 score to see how the performance curve works there (K6-2-400 had this reputation, I don't know how much of it is deserved, of being the point where diminishing returns start to take place for the K6-2).

Also kinda neat to take a look at the lower scores. Very interesting that a P55c-200 with 512K L2 smacks the P55c-233 with 256K L2 by more than ten percent.

Coppermine en Athlon resultaten zitten er nog niet bij, maar het is te verwachten dat een Coppermine het erg goed doet met z'n snelle L2 cache.

JC heeft een kort artikel gepost over het verhaaltje van een of andere stoere Japanner die z'n Athlon van DDR SRAM voorzien heeft. Dit systeem zou op gelijke kloksnelheid een 35% hogere CPUMark99 score geven als een normale Athlon, wat nogal onwaarschijnlijk klinkt:

According to here, K7-600 is 56.0, K7-650 is 60.0. According to here, K7-600 is 55.9, K7-650 is 57.8, K7-700 is 64.0, K7-800 is 69.4. A score of 95.3 would be boosting the Athlon's CPUMark by a factor greater than 35%, clock for clock. According to here, changing the 512K L2 on the PIII to 256K, then doubling the clock and better than halving the latency (among some other stuff) yielded an improvement of 24%.

The numbers here pressure us into believing that keeping Athlon's L2 at 512K, but only doubling the clock (not changing the latency) will result in performance increasing by a greater margin. Possible. I've heard that Athlon is hungrier for better caching, as it has more functional units and pipelines and suchlike, so maybe it'll get a greater than expected boost from faster caches. Nonetheless, this is one of those "too good to be true" deals -- If the additional cache improvements on the Thunderbird (lower latency, maybe higher width) still leave it "only" as fast as this Frankenthlon, then you'd find, umm, , a Tb-800 beating a PIII-1000 in this benchmark. I'm not quite ready to accept all this.

De jongens en meisjes van ComputerNerd hebben een eigen Athlon Gold Finger Device in elkaar geknutseld. De RAK71 POW!er Boost zal vanaf eind januari verkrijgbaar zijn en kost $44,95 voor de versie zonder L2 cache control en $59,95 voor de volledige versie:

ComputerNerd has designed the POW!er Boost controller with all the capabilities needed for FULL overclocking control, not just some of them. A prime objective of many overclockers is to take an inexpensive Athlon® rated at a lower speed to 750 MHz or even faster. To accomplish this, the divisor for the L2 cache onboard the processor has to be throttled back to a lower value and, unfortunately, this cannot be accomplished via the Golden Fingers edge connector on the CPU. However, it can be done using "passive" control in which resistors on the CPU are replaced by switched resistors in the control device wired to the same points. The POW!er Boost is the first and only active Golden Fingers controller announced to date with provisions for controlling the Athlon®'s L2 cache divisor. The card is designed with the switches and layout needed for switched cache control, while the resistors and cables are optional. The user can save money by buying the card without these components, but can upgrade by adding them himself or by trading up to the full-featured version.

A final basic objective of the POW!er Boost design is to offer the highest quality, greatest operational flexibility and absolute reliability with avoidance of frills that add to cost but not real value, The means that have been used to achieve these objectives include:

Use of a manufactured dual-sided printed circuit card with surface mounted resistors. With dimensions of only 1 x 2 inches, the board is the smallest controller yet announced (and close to the smallest possible). The small size lowers cost of the controller, makes it less obtrusive and reduces risk of accidental damage. It also increases reliability compared with hand-wired circuits. The POW!er Boost carries an unconditional lifetime warranty.

The compact dimensions allow usage with or without the CPU cartridge faceplate and with tall heatsinks such as ComputerNerd's EIGHTY-ON Athlon® cooler. Both features may be important where very effective cooling is needed, and some competitive products do not have these capabilities.

All cables (5 Volt power and L2 cache leads) are soldered to the board, eliminating the cost, unreliability and wasted space of connectors.

Power is taken from a pass-through power-supply connector and NOT from a 12-Volt motherboard header which would be a waste of power and dangerous for the motherboard.

Switches are conventional DIP type, NOT rotary. Use of rotary switches would add to cost and still require a look-up table to determine the required setting.

Users set the BP-FID multiplier with only four SPDT switches rather than having to set eight SPST switches as in some other products.

A LED power-on indicator light is NOT provided because it would add to cost and be of zero real value.

Price for the Basic version of the POW!er Boost controller (without L2 cache resistors and cables) is $44.95, and the Advanced version with full hardware for L2 cache control is $59.95. It is both the lowest-priced and most powerful active Golden Fingers controller yet announced. Deliveries will commence by the end of January.

Hans schrijft dat er een nieuwe BIOS update gereleased is voor de Biostar M7MKA Slot-A Athlon plank. De update fixed een UDMA66 detection error. Downloaden mag op deze page.

wildhagen schrijft: "De auteur van het welbekende CIH-virus (ook bekend als Tsjernobyl-virus) heeft een baan gekregen bij een Taiwanees bedrijf, om de beveiliging te testen ":

Taiwan technology firm keen to test its own hardware has hired the super-hacker who created the notorious Chernobyl virus -- which laid waste to hundreds of thousands of computers worldwide in April.

[...] "Our chairman felt he was a rare computer professional and we decided to accept him with an open heart," said Wahoo spokeswoman Vivi Wang.

Chen works in Wahoo's hardware testing department, she said.

Fiets richting ZDNET voor de rest van het artikel.

wildhagen schrijft: "Shit ! Toch geen paddo's in de lucht ":

MOSCOW--The first Russian nuclear reactor to cross into the new millennium survived the Y2K computer bug test, Russia's atomic power company said in Moscow today half an hour after the plant entered the New Year.

"At 3:00 p.m. Moscow time (1200 GMT) the staff of the Bilibinsk nuclear power plant met the New Year. In connection with the change of date no shutdown in the work of the equipment was observed," the Rosenergoatom firm, which manages eight of nine civilian nuclear power plants in Russia said in a press release at 1230 GMT.

Itar-Tass news agency said radiation levels at the plant remained within normal bounds.

The reactor, in Russia's far eastern Chukotka province opposite Alaska, was the first to cross into the New Year, just three hours after President Boris Yeltsin resigned in Moscow at noon Moscow time today.

An Emergencies Ministry duty officer in Moscow said there were no problems due to the Y2K computer bug as of 15 minutes after midnight in Chukotka.[break]Hebben we weer mazzel gehad . Verder meldt CNET ze in Australië ook geen last hebben gehad van eventuele Y2k problemen tijdens hun eeuwwisseling:[/break]CANBERRA--The Australia government said no Y2K computer bug problems had been reported as the country ticked over into 2000.

Senator Ian Campbell, parliamentary secretary to Australia's communications minister, told reporters shortly after midnight: "Certainly no failures have taken place, but I wouldn't expect any significant reports to occur until one o'clock or so."

Asked if the world could take heart from what happened here, he said, "It's looking good so far."

Campbell is stationed at the Emergency Management Australia office in the capital of Canberra to coordinate checks on Y2K problems across the country.

Asked about key sectors such as water, electricity, aviation, he said: "I've not had great concerns about those sectors, because I think they've done the work."

Movemoor schrijft: "Kennelijk zijn de anders zo preciese Zwitsers iets vergeten. In elk geval is het volgens hun in nieuw-zeeland nu 19100":

Bovenstaande heb ik van deze site geplukt. Movemoor was een van de 3 mensen die het nieuws had opgestuurd.

wildhagen komt vertellen dat Netscape 4.71 op Netscape's FTP is gezet. Hieronder de info van Beta Test:

I saw this over on MRTech this morning and thought some of you might want to grab it. I'm not sure what's been updated but it does have a cool new desktop icon.

Netscape 4.71 kun je hier naar beneden laden.

Beta Test heeft wat info over hoe je ICQ 2000 alpha nu wel aan de praat kunt krijgen. Hieronder de info:

Well, everybody is for ICQ 2000a - at least it seems like that. A lot of people asked us, how to install the latest version of ICQ. First of all, you need to know that ICQ 2000a has been restricted to a very small number of betatesters, who are allowed to test it. You also want to test the new software ? No problem, just follow the instructions:

1. Download ICQ 2000a from here.
2. Get our ICQ 2000a loader.
3. Save the file to the directory where you put ICQ 2000a and run it.
4. Make sure NOT to be connected to the net when starting the loader.
5. If you get an error message "Can't read process" don't give up too fast ! This is because the loader may not work properly with Windows NT and Windows 2000. Try to run the loader several times.

"Stick together team" zegt een zachte stem door je hoofdtelefoon, en al spoedig hoor je links en rechts "roger that" langzaam kruip je naar een gat in de grond. Nog even en je betreedt een stuk riool... "Take Position, and wait for my go" zegt iemand die kennelijk de leiding maar op zich heeft genomen... "fire in the hull!!" schreeuwt iemand die zijn zenuwen niet helemaal meer in bedwang heeft, en flikkert een flash granade naar beneden. De grenaat komt tot ontploffing, en met een ferm "Go Go Go" duiken jullie allemaal het gat in. Hell breekt los, links hoor je het karakteristieke geblaf van een Desert Eagle, een pistool, rechts zwiept een grote dolk, wat verderweg ligt iemand met een G3/SG-1 Sniper Rifle een teammate omver te knallen. Dan zie je een gewillig doelwit, en de ammoband van je FN M249 PARA begint in beweging te komen... uiteindelijk hoor je een doffe knal en zie je de vloer razend snel op je af komen. Iemand was omgelopen en heeft van korte afstand een schot hagel met een M3 Super 90 Combat in je rug geschoten - Counterstrike is een spel voor tacttics, niet de bedoeling om in je eentje een potje rond te rausen.

Als je het spel door hebt, je teamgenoten dekt en je doel berijkt staat je geld te wachten voor steeds mooiere wapens.

The time has come. The CS Team, Barking Dog Studios and Valve Software are proud to present Counter-Strike Beta 5.0!! Head over to the downloads page to start downloading. There are 2 files for windows users, a beta4 to beta5 upgrade (12 megs) and a full beta5 installation (31 megs). The linux port is also available for linux server admins.

Note to Server OPS : DO NOT RUN CS_ASSAULT. There's a bug in our code that will crash the server whenever someone blows up the satellite dish.

Get your act together and pull yourself into rank!

Download Counter Strike 5 en kom mee doen, bijvoorbeeld op onze Clan Server! (130.161.58.110)

Ik kwam via Slashdot bij Uptime.net, waar je wedstrijdje uptime kunt spelen. De hoogste uptime is op dit moment in handen van een FreeBSD systeem, die al 1994 dagen onafgebroken staat te pruttelen. Win98 doet het ook niet slecht met een maximale uptime van 61 dagen (zolang je het ding niet aanraakt blijft de tent meestal wel draaien ):

Uptime is a program that keeps track of your server uptime, and lets you compare it with the uptime of other (similar) hosts. It consists of two parts, a client- and a serverpart. The client runs on your server and sends the uptime every few minutes to the server. The server collects all data in a table on this site.

Uptimes per OS (avg, max)
NetBSD	147 days, 03:39m	1322 days, 04:17m
FreeBSD	130 days, 00:34m	1994 days, 20:39m
OpenBSD	115 days, 16:38m	395 days, 19:24m
Other	96 days, 07:22m	769 days, 22:37m
Linux	68 days, 18:24m	994 days, 04:52m
Novell	66 days, 06:51m	423 days, 20:05m
Solaris	45 days, 20:28m	149 days, 20:37m
Windows NT	12 days, 19:14m	49 days, 17:02m
BSDi	10 days, 20:23m	279 days, 13:00m
Windows 95	6 days, 16:42m	49 days, 17:02m
Windows 2000	4 days, 20:35m	49 days, 17:02m
BeOS	4 days, 01:35m	35 days, 11:24m
Windows 98	2 days, 19:36m	61 days, 13:52m
Multiple	1 day, 22:31m	100 days, 01:03m

Onze eigen server zit op dit moment aan 16 dagen .

HardOCP heeft een artikeltje gebouwd rond hun overgeklokte PIII-650, die dankzij peltier koeling stabiel draait op 806MHz.

I had to go Peltier. Pretty lightweight though. 33W, 15V Peltier running at only 7 volts. No below zero temperatures, no condensation, no insulation. Just had the chip, a small 1 inch square plate of aluminum, the 1 inch square Peltier and the stock heatsink, all sandwiched & joined together.

Here is the peltier sandwich and a close-up:

Alpha heeft een speciale versie van z'n Athlon heatsink gemaakt voor slot-A moederborden die hun power connector ongelukkig voor de processor geplaatst hebben. Dit geeft ruzie wanneer je een grote heatsink aan wilt leggen, daarom heeft Alpha in de P7125 heatsink een uitsparing aangebracht:

Ultiem leedvermaak voor de tweaker (met brute, supersnelle & stabiele homemade PC): News.com bericht dat gebruikers van Comprak Presario 5800 en 5900 PC's massaal klagen over weglekkende resources in Win98:

Compaq Computer today attempted to snuff out a problem with some of its consumer PCs that is caused by these systems' interaction with Microsoft's Windows 98 operating system.

Many customers have been complaining that Compaq's Presario 5800 and 5900 models have been freezing up. The complaints increased around Christmas, and the computer maker's customer support operations--both telephone and online--have been swamped dealing with the problem, Compaq said.

Too many software programs, including Windows 98, are competing for limited resources, causing the systems to slow to a crawl, Compaq and analysts said.

3D Alpha keek naar de performance van de Cheetah 18LP en Barracuda 18LP, twee errug dikke U2W SCSI drives van Seagate. Hier een hap uit de review:

So I decided to see how fast programs would load and operate. Though I couldn't stopwatch something like this very accurately, I can just say that if you want something done: This drive will run like hell to get it for you. Loading up large memory consuming applications such as Netscape are absolutely no problem with this drive. In fact, unlike your old UDMA/33 drives, you may find that you'll have to eliminate that donut and coffee break during a load session of a game or Word97. The easiest way to understand what I'm trying to describe is to either try it yourself, or just imagine having programs load without any sort of slowdown whatsoever. Sounds pretty outlandish, doesn't it? The scary thing is that there are faster drives than Ultra2 (Ultra160 for example.), and they boast faster transfers and lower seek times, but you can't deny that something like this is nothing other than fast defined.

Onze Russische vrienden van iXBT Labs hebben een review gepost van de Supermicro PIIISCD, een i820 plank met 2 DIMM slots. SuperMicro - bekend van z'n onoverklokbare server borden - blijkt voor de verandering ook overclocking features toegevoegd te hebben:

First of all, the board has a special jumper for manual resetting of the external frequency. The available values are: AUTO, 100 and 133MHz. AUTO also allows you to get 66MHz with Intel Celeron or old 66MHz Pentium II. Secondly, BIOS Setup of this mainboard supports an even broader frequency range: 66, 100, 105, 114, 120, 124, 128, 133, 138, 143, 148, 150, 152, 150 and 160MHz. On the other hand however, overclocking arouses some serious problems. And again as in many other cases it is Intel that hatches the mischief. FWH from the i820 chipset is clocked depending on the PCI bus frequency. But if you set the FSB frequency other than 66, 100 or 133MHz, the PCI bus frequency appears a bit higher than the required one. And since this hub doesn't prove very stable at higher frequencies, 124 and 128MHz FSB appear simply useless: the mainboard doesn't work at all. And as for the opportunity to set the PCI divider manually, Supermicro P3SCD doesn't have any, unfortunately. So, in general, if you want to overclock your CPU, you'd better do it on the mainboards by some other manufacturers such as the beloved ABIT, for instance, although it is possible to do it on Supermicro as well.

AnandTech heeft een review van de AOpen AX6BC Pro II Millennium Edition geplaatst. Een speciaal plankje dat vanwege z'n goede stabiliteit werd aangeprezen als de AnandTech Editor's Choice:

Take the only motherboard to receive AnandTech's Gold Editor's Choice Award in a standalone review, the AOpen AX6BC Pro Gold, make some basic, but useful improvements, and you get the AX6BC Pro II Millennium Edition - the best overall i440BX motherboard on the market.

For the hardcore overclocker, the latest offerings from ABIT, the BE6-2 and BF6 may be a better solution thanks to SoftMenu III's increased tweaking flexibility. For everyone else, there is no reason to pass on the AX6BC Pro II and its rock solid stability under all conditions. That is, of course, if you still want an i440BX board this late in its life cycle...

Jannis schrijft: "Kwam op Linux Today een artikel tegen over een petitie van Libranet, die eist dat hardwarefabrikanten meer Linux drivers uitbrengen. De organisatie hoopt minimaal 2.000.000 handtekeningen binnen te halen:":

"Libranet Announces 'The Linux Drivers Petition'." "Libra Computer Systems, producer of the Linux by Libranet distribution, today announced the beginning of a campaign to help bring more pc drivers to Linux."

"Linux developers currently do an excellent job of producing pc device drivers often working in the absence of any specifications. They could do an even better job if they had the co-operation of the manufacturers."

Meer info over de petitie vind je in de press release. Meedoen kan hier.

Dat is zo'n beetje de strekking van het artikel dat FPS3D.com heeft geschreven over de Voodoo4 en Voodoo5 videokaarten, die 3dfx het komend voorjaar zal introduceren. Met een fillrate van 1333Mtex/sec zal de Voodoo5 6000 ongetwijfeld nieuwe snelheidsrecords zetten, het is alleen de vraag of dit een prijs van 600 dollar rechtvaardigd:

The most important downfall has to be the lack of Double Data Rate memory support. By the time the 3dfx cards are released, nVIDIA will have unleashed their 64mb DDR GeForce unto the public. 32mb DDR will then become an affordable standard, and where is 3dfx left then? You'll have a choice between their 128mb of SDRAM for $600.. or nVIDIA's offer of 64mb of higher-powered DDR SDRAM for what is being speculated at $350 and under.

This is where 3dfx's usual onslaught of fill rate claims are defeated. We have already seen the alarming increase in framerate on 32mb GeForce cards in DDR format as opposed to their vanilla SDRAM and/or SGRAM counterparts [ed: this is because larger memory bandwidth allows for 32-bit color without anywhere near as much of a performance hit. It also increases performance in games with high-resolution textures.]. I think it's pretty safe to say, at this rate, the 64mb DDR GeForce should certainly be powerful enough to at least rival the second-highest tier on the V5 series (Voodoo 5 5500 - $300), which boasts around 667-733 Megapixels per second [ed: and also we can basically count on the fact that nVidia will have increased clock speeds by then or soon after.].

If this is the case, 3dfx will be left with nothing to combat the current champion. GeForce boasts hardware transform & lighting, processor independent rendering, equivalent fillrate, lower prices, better availability, TV-OUT, multiple manufacturers (so we can have some nice price slashing due to competition - in addition to fun features like 3D Glasses, tweaked out drivers, etc.), and an overall superior reputation (due to the bad taste the V3 left in the mouths of some of its users).

Sharky Extreme heeft een artikel gepost waarin ze dieper ingraven op de performance en beeldkwaliteit van de Savage2000. De Savage2K blijkt nogal vreemde renderfouten te maken, zoals deze zaagtand effecten in Quake III Arena (links GeForce, rechts Savage2000):

Glorix Jim komt vertellen dat Storm Linux 2000, een op Debian gebaseerde Linux distri, officieel gereleased is: "Wordt Linux dan misschien een echt een beetje gebruikersvriendelijk voor netwerkbeheerders? ":

It's Here! The download edition of Storm Linux has been released and is available now for download. Featuring a graphical lilo/boot manager, the Storm Package Manager, a new user friendly GUI for installing, updating and removing Debian packages, we think you'll find Storm Linux 2000 the easiest, most flexible way to get the power and freedom of Debian GNU/Linux.

Storm Administration System

At the heart of Storm Linux is the Storm Administration System (SAS), a set of tools that make Linux simpler to set-up and configure. New Linux users will find SAS easy to use, but it's advanced features make it a powerful tool in the hands of a system administrator. With SAS, the network system administrator now has the choice of administering a network of computers remotely using the traditional command line or through a point and click interface.

Flip Zemanova stuurde een link naar dit artikel van NRC Handelsblad, over de maatregelen die bedrijven hebben getroffen als voorbereiding op het komende armaged...euh millennium: "Water, licht, verwarming, eten.... wat maakt het uit als ik maar kan internetten kom ik de 1e week wel door ":

De burger hoeft zich geen zorgen te maken over de millenniumbug, stelt de overheid. Bedrijven hebben wel maatregelen getroffen.

[...] Talloze bedrijven en instituten hebben zich wel op alle eventualiteiten voorbereid. Zij hebben water- en dieselvoorraden aangelegd, namen extra grondstoffen en reserve-onderdelen in huis en installeerden telefoonsystemen die onafhankelijk werken van de KPN. Zelfs staan boodschappers met fietsen en motorfietsen klaar. Op diverse plaatsen koppelen bedrijven zich op voorhand los van het elektriciteitsnet. Onbemande pompstations zijn bemand. Polderbesturen verlagen het polderpeil, luchtvaartmaatschappijen halen vliegtuigen terug. De overheid heeft begrip en waardering uitgesproken voor de maatregelen: eindelijk heeft het bedrijfsleven zijn noodplannen op orde. Niemand zal honen als de maatregelen uiteindelijk overbodig blijken.

Nog een paar uur wachten en dan weten we of de pleuris uit breekt in Nieuw Zeeland en Australië .

Volgens Daiki is er een proggie in de maak waarmee de L2 cache divider van de Athlon softwarematig getweakt kan worden:

We already know we can change Athlon's voltage and multiplier without any soldering. But what about L2 divider?

OK, we will acquire Athlon's L2 tweaking utility soon, which can change Athlon's L2 through software. This means that we no more need soldering to change L2 divider. Sounds Good ? [break] Verder heeft-ie ook Distributed.net CSC resultaten van een Athlon op 835MHz met DDR SRAM: [/break] I'm now keeping to contact awesome man who modified Athlon500(week36)'s L2. This man replaced his L2 cache with 2.1ns engineering sample SRAM.

This engineering sample SRAM runs with xor, so this modified Athlon runs with full speed DDR-SRAM L2 cache though it has still very high latency. This great guy is testing several benchmarks now, and I confirmed one of them.

CSC Keyrate(CSC Distributed Client version 1.04): 199231 Kbps

Yeah, DDR Full Speed Athlon835MHz was able to get approximately 200000 Kbps !! Do you know how fast it is? You can download this client from Dcypher.net and test it on your CPU.