I was cursed with a Savage4 after having a Voodoo 3 for ages... it quickly became known in our household as the Savage Whore, for the sheer number of times it got fucked one way or another.
The drivers were absymal, the hardware regularly locked up and it got replaced real fast.
If the branches are local, yes of course they can, that's the whole point of the dedicated branch prediction part of the pipeline.
The point was that if the area you're branching into is a cache miss, you lose a whole bunch of cycles which cannot easily be caught up.
If we assume it's a cache hit, due to the size of the pipeline, you gain a significant amount of performance.
As an aside, compilers do not optimise kernel functions inline into dynamically loadable modules, e.g the filesystem drivers we were discussing a few GPs up. This will online occur if they are explicitly inlined.
Have a think about what keeping code very local may do for you, specifically, bear in mind the size of the cache lines and length of the pipeline in today's processors.
The real issue though is that often developers skirt around the Win32 API spec and end up "doing it wrong" by talking directly to under-the-hood parts of the operating system, especially where copy protection is concerned.
No-one seems to worry about this much, as once you've tested with Vista and XP now, you're largely set for gaming customers.
So the end result, is no-one worries about adhering to the API specs as there is no commercial driving reason to do so.
Presumably you understand that most of the OS type guessing in nmap uses quirks in the TCP stacks to determine which OS it is.
So if the "first" box you talk to exhibits FreeBSD quirks, ie the load balancer / cache, no matter what is behind it, it will be identified as FreeBSD, right?
It's actually the other way around - they used to show up in Netcraft as Linux servers even though they were IIS on Windows Server 2003 for a long time.
This is because the server version reported was actually Akamai's balancing and caching infrastructure in front of the Hotmail servers.
This is the problem really, I think we need to re-draw of what percentage of junk output constitutes a "bad" ISP which should be cut off upstream.
I understand there's a massive cost to implement monitoring, but that has to be part and parcel of remaining connected.
You would think at least the US, western Europe and Korea could get this sorted though, then we can pressure China, Russia and all the rest to follow suit.
It notices nasty things happening such as SSH scanning and mass SMTP sessions.
First they try to contact you to confirm whether it is your activity or a zombie, if they cannot get hold of you, they pull the plug if it is severe enough.
I liked it - I got a notification when I lost control of an old Linux box on my network before I came home and spotted it.
This should really always be part of the solution, being adequately monitored by your upstream provider.
Yes I often buy on Steam too, the fact it seems to overcome the "losing the media" hurdle helps too.
I do resent the prices usually being higher than the store price though! Usually as you say, it is worth the hassle of not going to the store. Recent titles have been up to a week late onto Steam though.
Also the recent case of GTA IV dropping to £18 just two months after release is a bit sickening.
I also agree on pointing the finger firmly at EA - they one of the biggest supporters of DRM even after the backlash they have suffered.
For me it's a pain, as a few of the titles I play do come from EA, eg the Battlefield series.
I call BS on this. Piracy gets people interested in both games and media, as recent news articles about studies of the subject seem to indicate.
The real effect of DRM on me is that each time I buy movies or games I am discouraged from doing so again and pushed a little further towards so-called piracy. I'll explain why...
As a teenager, I was prolific pirate. Encouraged at first by how easy it was, and the fact I really did not have the means to purchase the games. On rare occasions, I was a customer, but only when funded by birthdays etc. I was never a potential customer for the games I pirated.
Now, I'm an adult with cash I am happy to spend on games. However, it was my teenage years which got me hooked on gaming.
What makes me sad is that I still find often find myself resorting to piracy. The driving factors:
1) I do not like waiting to play games other regions have, I often download US or Japanese games before they reach Europe. The difference is that now I will happily buy it when it (finally) reaches our European shores.
2) Copy protection - I don't like physical media, it gets scratched, I lose it and do not like switching DVDs all the time. I've been buying tons of Steam games lately for this very reason. This annoys me, as I have much less rights with a Steam copy of a game - e.g. no chance of reselling it. I had hoped Steam and other digital distribution would stop the region-delays game too... but it has not. The activation limit policies on newer PC games are also starting to cause this.
3) Same crap we get with DVDs loaded with no-skip trailers and anti piracy warnings. The DVD rips are conveniently pre-cleaned of the BS they force in my face on a legal copy. Again, the region delays suck too.
Having pirated this stuff, I am often willing to buy a copy when it finally reaches Europe. It worries me that by going about it this way I am risking finding myself in court for copyright infringement, but honestly, I am happy to pay for the content. I would just really prefer it is delivered sensibly without the stupid limitations listed above.
The result of this is that when I finally buy a copy, each time I am disappointed to find it significantly less convenient than the pirated copy, a little bit of my willingness to continue buying is forever sucked into the void.
I have the PBSCV service installed, running correctly, the game connecting to it, PnkBstrA communicating with the PB servers, and full admin rights.
The issue is that PB kicks you for the API calls it does not recognise, as I explained in another post above.
It assumes any non-whitelisted API calls are hack tools doing their work, so EvenBalance either need to update it fully, or disable the API monitoring when a W7 system is detected for now.
I can imagine them not wanting to do the latter as this would compromise security.
On the other hand, the long-term private hacks available for BF2 seem to have no problem hiding from PB.
I think the crux of this is that for all the games, a whitelist of "allowed" API calls exists.
It would seem a lot of the W7 APIs are not on this list, or because all the addresses are now different, they are not recognised.
If it's the latter, I'm guessing they want the API to be stable before doing the hard work of whitelisting each API address. They might be working on it quietly, just refusing to promise anything until they have results.
I like this approach as it only ever leads to nice surprises for consumers.
For me, the single biggest problem with Windows 7 gaming is the lack of PunkBuster support, as EvenBalance are refusing to support the beta at this stage.
Fair enough, it is their choice, however the beta is public, so many of their customers are in the same boat here. It seems sad that the public statements seem to indicate they are not even willing to look at it.
By throwing money at Alzheimers reserach and on a more personal level by doing all he can to remain as mentally active as possible.
It is worth noting that many believe the cause is now identified and that large pharma companies are working on getting treatments through trials.
I seem to recall a headline on/. linking high blood sugar to the memory loss effect. Hopefully this shows that more research is shedding more light on the condition.
I'm another in agreement with this - but will say I still do not have passwords enabled and solely use key based authentication for SSH.
I understand the argument against security through obscurity, but also believe that given a simple opportuniyu to decrease your attack surface it should definitely be taken!
Confession... both of the above changes were made after a really stupid day giving my GF access to my music via Samba lead to me creating an account with the username "mp3", the password "mp3" and forgetting that it would by default have an SSH enabled account. The machine was scanned and brute forced within about 3 days (surprised it took that long TBH).
sshblack is also really against stupid attackers (ie anyone except the distributed attackers in the articule).
Yeah that was actually one of the deciding factors for me too - I have an older (around 2000) motherboard from them which has caps that are bloating at the top.
Once Asus and Gigabyte went all solid state I think the other manufacturers in the mid-range had to follow suit.
I was pleased to see it on the Abit - it was definitely a "top of the mid-range" board, so was not something I automatically expected it to have.
On the other hand I have a friend who always buys cheap components, cheap motherboard at the top of the current AMD cycle, spends about half as much and accepts the fact it might just break. Assuming less than half of the systems break, cost wise, he's doing better. Definitely not something to be doing for high reliability though.
Ah yes, there is an issue with that - when JEDEC specified the DDR2 specs, the SPD info on the RAM can only go up to 800 MHz, so in the official spec, there is no way for newer DIMMs to indicate that they are 1066 MHz modules.
This has been overcome to some extent - nVidia have created an unofficial extension to the SPD specification, called EPP. The problem is that manufacturer support for this is very patchy.
Almost all newer nForce chipsets boards support it, but amongst other manufacturers it is very hit or miss. Most likely on an Intel chipset board, you will have to set the RAM speed and the timings yourself.
I've found Corsair's forums quite useful - there's a lot of posts with tights timings for the Dominator RAM on most motherboards.
Oops I lied.. there's also XMS2 Dominator, which IS what I have. You can tell cos there's a D on the end and big fat heatsinks. The page for that is here...
Erm, I have XMS2 Dominator 8500C5, which is guaranteed up to 1066, however it will only do this above the stock 1.8V, it is intended to be used at 2.1V. There are different XMS2s though, check Corsair's website for the spec on the exact model number. http://www.corsair.com/products/xms2/default.aspx
Originally I had to drop the multiplier on my CPU to achieve this, but now have got it stable with the multiplier intended for 800MHz ram by upping the core voltage a little (1.4V instead of 1.35V) and bumping most other ICH9 voltages by 0.1V, then putting a better fan on it.
Slashdot Mirror
So your cat agreed to the EULA, and by doing so, gained a license to use the software, for themselves.
So you still have no license to use it...
The fact that the software is now installed on your PC, does not mean it is yours. You might as well torrent it.
Where's the big news?
I think AS47868 was causing the routes to be lost.
It was making mass BGP announcements about really long incorrect routes.
I was cursed with a Savage4 after having a Voodoo 3 for ages... it quickly became known in our household as the Savage Whore, for the sheer number of times it got fucked one way or another.
The drivers were absymal, the hardware regularly locked up and it got replaced real fast.
oops, typo... s/online/only/
If the branches are local, yes of course they can, that's the whole point of the dedicated branch prediction part of the pipeline.
The point was that if the area you're branching into is a cache miss, you lose a whole bunch of cycles which cannot easily be caught up.
If we assume it's a cache hit, due to the size of the pipeline, you gain a significant amount of performance.
As an aside, compilers do not optimise kernel functions inline into dynamically loadable modules, e.g the filesystem drivers we were discussing a few GPs up. This will online occur if they are explicitly inlined.
Have a think about what keeping code very local may do for you, specifically, bear in mind the size of the cache lines and length of the pipeline in today's processors.
The real issue though is that often developers skirt around the Win32 API spec and end up "doing it wrong" by talking directly to under-the-hood parts of the operating system, especially where copy protection is concerned.
No-one seems to worry about this much, as once you've tested with Vista and XP now, you're largely set for gaming customers.
So the end result, is no-one worries about adhering to the API specs as there is no commercial driving reason to do so.
Presumably you understand that most of the OS type guessing in nmap uses quirks in the TCP stacks to determine which OS it is.
So if the "first" box you talk to exhibits FreeBSD quirks, ie the load balancer / cache, no matter what is behind it, it will be identified as FreeBSD, right?
It's actually the other way around - they used to show up in Netcraft as Linux servers even though they were IIS on Windows Server 2003 for a long time.
This is because the server version reported was actually Akamai's balancing and caching infrastructure in front of the Hotmail servers.
This is the problem really, I think we need to re-draw of what percentage of junk output constitutes a "bad" ISP which should be cut off upstream.
I understand there's a massive cost to implement monitoring, but that has to be part and parcel of remaining connected.
You would think at least the US, western Europe and Korea could get this sorted though, then we can pressure China, Russia and all the rest to follow suit.
Here my ISP has an IDS watching user activity.
It notices nasty things happening such as SSH scanning and mass SMTP sessions.
First they try to contact you to confirm whether it is your activity or a zombie, if they cannot get hold of you, they pull the plug if it is severe enough.
I liked it - I got a notification when I lost control of an old Linux box on my network before I came home and spotted it.
This should really always be part of the solution, being adequately monitored by your upstream provider.
Yes I often buy on Steam too, the fact it seems to overcome the "losing the media" hurdle helps too.
I do resent the prices usually being higher than the store price though! Usually as you say, it is worth the hassle of not going to the store. Recent titles have been up to a week late onto Steam though.
Also the recent case of GTA IV dropping to £18 just two months after release is a bit sickening.
I also agree on pointing the finger firmly at EA - they one of the biggest supporters of DRM even after the backlash they have suffered.
For me it's a pain, as a few of the titles I play do come from EA, eg the Battlefield series.
I call BS on this. Piracy gets people interested in both games and media, as recent news articles about studies of the subject seem to indicate.
The real effect of DRM on me is that each time I buy movies or games I am discouraged from doing so again and pushed a little further towards so-called piracy. I'll explain why...
As a teenager, I was prolific pirate.
Encouraged at first by how easy it was, and the fact I really did not have the means to purchase the games.
On rare occasions, I was a customer, but only when funded by birthdays etc.
I was never a potential customer for the games I pirated.
Now, I'm an adult with cash I am happy to spend on games. However, it was my teenage years which got me hooked on gaming.
What makes me sad is that I still find often find myself resorting to piracy.
The driving factors:
1) I do not like waiting to play games other regions have, I often download US or Japanese games before they reach Europe.
The difference is that now I will happily buy it when it (finally) reaches our European shores.
2) Copy protection - I don't like physical media, it gets scratched, I lose it and do not like switching DVDs all the time. I've been buying tons of Steam games lately for this very reason.
This annoys me, as I have much less rights with a Steam copy of a game - e.g. no chance of reselling it. I had hoped Steam and other digital distribution would stop the region-delays game too... but it has not.
The activation limit policies on newer PC games are also starting to cause this.
3) Same crap we get with DVDs loaded with no-skip trailers and anti piracy warnings. The DVD rips are conveniently pre-cleaned of the BS they force in my face on a legal copy. Again, the region delays suck too.
Having pirated this stuff, I am often willing to buy a copy when it finally reaches Europe.
It worries me that by going about it this way I am risking finding myself in court for copyright infringement, but honestly, I am happy to pay for the content. I would just really prefer it is delivered sensibly without the stupid limitations listed above.
The result of this is that when I finally buy a copy, each time I am disappointed to find it significantly less convenient than the pirated copy, a little bit of my willingness to continue buying is forever sucked into the void.
I have the PBSCV service installed, running correctly, the game connecting to it, PnkBstrA communicating with the PB servers, and full admin rights.
The issue is that PB kicks you for the API calls it does not recognise, as I explained in another post above.
It assumes any non-whitelisted API calls are hack tools doing their work, so EvenBalance either need to update it fully, or disable the API monitoring when a W7 system is detected for now.
I can imagine them not wanting to do the latter as this would compromise security.
On the other hand, the long-term private hacks available for BF2 seem to have no problem hiding from PB.
I think the crux of this is that for all the games, a whitelist of "allowed" API calls exists.
It would seem a lot of the W7 APIs are not on this list, or because all the addresses are now different, they are not recognised.
If it's the latter, I'm guessing they want the API to be stable before doing the hard work of whitelisting each API address. They might be working on it quietly, just refusing to promise anything until they have results.
I like this approach as it only ever leads to nice surprises for consumers.
Maybe it's optimistic moaning... the point of a beta is to get feedback.
With no feedback, nothing gets fixed.
The more vocal people are about their issues, the more likely they are to be fixed.
I use all of the games I tested on exactly that platform - Vista x64 with over 4GB RAM.
I know all of the following work perfectly, as I play them all BF2, BF2142, COD:WAW, COD4, Far Cry 2, Crysis and ET:QW.
So yes, I AM blaming the beta.
For me, the single biggest problem with Windows 7 gaming is the lack of PunkBuster support, as EvenBalance are refusing to support the beta at this stage.
Fair enough, it is their choice, however the beta is public, so many of their customers are in the same boat here. It seems sad that the public statements seem to indicate they are not even willing to look at it.
By throwing money at Alzheimers reserach and on a more personal level by doing all he can to remain as mentally active as possible.
It is worth noting that many believe the cause is now identified and that large pharma companies are working on getting treatments through trials.
I seem to recall a headline on /. linking high blood sugar to the memory loss effect. Hopefully this shows that more research is shedding more light on the condition.
He said gaming, not retro-gaming.
SM2.0? Welcome to five years ago.
I'm another in agreement with this - but will say I still do not have passwords enabled and solely use key based authentication for SSH.
I understand the argument against security through obscurity, but also believe that given a simple opportuniyu to decrease your attack surface it should definitely be taken!
Confession... both of the above changes were made after a really stupid day giving my GF access to my music via Samba lead to me creating an account with the username "mp3", the password "mp3" and forgetting that it would by default have an SSH enabled account. The machine was scanned and brute forced within about 3 days (surprised it took that long TBH).
sshblack is also really against stupid attackers (ie anyone except the distributed attackers in the articule).
Yeah that was actually one of the deciding factors for me too - I have an older (around 2000) motherboard from them which has caps that are bloating at the top.
Once Asus and Gigabyte went all solid state I think the other manufacturers in the mid-range had to follow suit.
I was pleased to see it on the Abit - it was definitely a "top of the mid-range" board, so was not something I automatically expected it to have.
On the other hand I have a friend who always buys cheap components, cheap motherboard at the top of the current AMD cycle, spends about half as much and accepts the fact it might just break.
Assuming less than half of the systems break, cost wise, he's doing better. Definitely not something to be doing for high reliability though.
Ah yes, there is an issue with that - when JEDEC specified the DDR2 specs, the SPD info on the RAM can only go up to 800 MHz, so in the official spec, there is no way for newer DIMMs to indicate that they are 1066 MHz modules.
This has been overcome to some extent - nVidia have created an unofficial extension to the SPD specification, called EPP. The problem is that manufacturer support for this is very patchy.
Almost all newer nForce chipsets boards support it, but amongst other manufacturers it is very hit or miss. Most likely on an Intel chipset board, you will have to set the RAM speed and the timings yourself.
I've found Corsair's forums quite useful - there's a lot of posts with tights timings for the Dominator RAM on most motherboards.
Oops I lied.. there's also XMS2 Dominator, which IS what I have. You can tell cos there's a D on the end and big fat heatsinks. The page for that is here...
http://www.corsair.com/products/dominator/default.aspx
Erm, I have XMS2 Dominator 8500C5, which is guaranteed up to 1066, however it will only do this above the stock 1.8V, it is intended to be used at 2.1V. There are different XMS2s though, check Corsair's website for the spec on the exact model number. http://www.corsair.com/products/xms2/default.aspx
Originally I had to drop the multiplier on my CPU to achieve this, but now have got it stable with the multiplier intended for 800MHz ram by upping the core voltage a little (1.4V instead of 1.35V) and bumping most other ICH9 voltages by 0.1V, then putting a better fan on it.