Can the intellivision use a light gun?

[8bitwidgets.com]

I pose this question to those who really know the Intellivision hardware.  I would be interested in trying to make one but before I dig into it, I wonder if it's even possible.

 

I'm familiar with the basic principles of how it works for the NES..  so i'm wondering if the NES had some special next-gen chip / capability..   but then i look back at the Odyssey and it was a light gun capable system before the Intellivision existed so clearly the technology already existed..  

 

That said, I know some hardware just isn't made to "talk" to a light gun the way the NES and Odyssey do.  Even the Odyssey 2 from what I've looked into it, is incapable of working with a light gun because of the way the chip was designed.  

 

So, are we going to be making a light gun for Goose Hunt?!

[+fdr4prez]

 

[8bitwidgets.com]

Ok let me narrow it down because that thread was all over the place..   basic parameters:

 

1. A light gun that works with a CRT.  Strictly following the NES approach.  (No nonsense of Wii and LCD approaches)

 

2. Making something custom to meet the pinout of the INTV.  A gun made from the ground up to work with the INTV hardware

 

3. Alternatively if the requirements of an existing gun could work with the INTV, make an adapter to re-route the signals needed by the gun to function.  

 

 

 

Can a stock intellivision be capable of reading the scan line signal with the level of accuracy that other light gun capable systems have?  

 

 

 

 

[8bitwidgets.com]

@decle weighed in with a message to me 30min ago..  and it sounds like the answer is yes we can achieve it with a stock INTV using a CRT.  The capability is there and that's all I'm after.  I have a CRT I use for gaming and would test against that.

 

The next big question is figuring out if there is an existing gun we can route the right signals too from a INTV port.  This is still stuff I really know little about.. like does the INTV even have the means of sending a voltage through the pinout to power the light gun flash and electronics?  

I'm willing to explore this with my partner who I believe would find the idea of getting a light gun to work interesting as well, but it's going to take some research to figure out the answers.  

 

 

As mentioned in the other thread, the Sega zapper would likely be the best starting point since it has a db9 port.. just generally easier to work with..  however making a DB9 to NES port isn't out of the question.  The parts can still be found / bought online.  

Edited March 20 by 8bitwidgets.com

[decle]

Hey all,

 

I believe NES Zapper / Sega Light Phaser work in a pretty simple way.  The gun has a photodiode and some electronics in it.  When you pull the trigger the game detects this and paints the screen black on the next frame.  The photodiode assesses the brightness at this point as a reference for ambient light.  Then on the following frame the game draws one or more white squares on the screen where the duck(s) were.  If the photodiode now registers more light than the reference it captured a frame ago a hit is registered.  There's a bit of faffing about for which of multiple targets are hit, but that's essentially the jist of it, as explained here:

 

 

The only small complication I immediately see to interfacing such a gun to the Intellivision II, SSVA or ECS is that the Inty controller ports don't supply power, and therefore, cannot run the electronics in the gun.  This could be solved by having a battery in an adapter circuit sitting between the gun and the Master Component.  There are plenty of adapter circuits online to interface these guns with other consoles that could provide a jumping off point for such a design.

 

It should also be noted that the NES and Sega guns don't play nicely with modern TVs.  It's not clear to me if this is just down to lag between the console and screen (which might be fixed in new games with a calibration screen and code to expect the lag), or something more fundamental.  Either which way, Inty zapper games would have to be specifically written to take advantage of the gun.  No just plugging in a phaser and blasting away at Space Battle.

 

Edited March 20 by decle

[8bitwidgets.com]

12 minutes ago, decle said:

or something more fundamental

yea this is only going to be made to work with CRT..  from my understanding is the very nature of how a CRT refreshes its screen is different than how a modern TV does it.  old tvs would do a full scan from top to bottom 60 times a second..  even a LCD monitor that is at 60hz doesn't quite do it the way it's needed to for the diode to pick up the flash..  

There is that modern LCD friendly light gun that works really well, but it's entirely new technology.  There is no way for a zapper to work with an LCD TV..

 

 

Also, thanks for confirming the voltage situation from the INTV..  that was a real concern I wondered about.  However yes a battery could fix that.  It's not like the zappers consume massive amounts of power.  So that could be addressed easy enough.  I'm not sure how soon we'll get to it, but I do want to find out more about the pinout of the sega zapper first.. and how different is it from the NES zapper.  maybe they are basically the same with different pin configurations?   

Ideally I would first look into some kind of adapter like the recent INTV to Flashback adapter.  The battery could be in there as well..  this way we don't have to go and engineer an entirely new light gun / trigger system.. but leverage existing easy to find / affordable classic guns.  


We have to get a bit further with the Coleco, 2600, and upcoming INTV MCPs to produce and release, plus making a new INTV Pong-like paddle, another goal for 2024.   The light gun will definitely be something I'd like to work on though or rather an adapter for an existing gun if possible first.. 

 

 

Thanks,

Caleb

[+Black_Tiger]

The Zapper sucks and the Light Phaser works great.

 

If a classic light gun is targeted, please use the Light Phaser.

[8bitwidgets.com]

50 minutes ago, Black_Tiger said:

The Zapper sucks

 

lol, but i think it's great, very accurate, lasts forever, but all nerd raging aside, I would actually aim to support both..  i can't help but imagine they likely function the same way just with different pins..  

I just found this page which is very interesting.. a light phaser to NES adapter.. so this is very encouraging as well that if we could get the zapper to work..  from a quick ebay search it seems like currently prices are pretty comparable for either one so this is good as well..  roughly $25-35 for either one..  some more some just but this was a quick average.. 

 

https://www.boojakascha.ch/index.php?page=phaser2nes

 

 

 

[+DZ-Jay]

On 3/20/2024 at 8:00 AM, decle said:

Hey all,

 

I believe NES Zapper / Sega Light Phaser work in a pretty simple way.  The gun has a photodiode and some electronics in it.  When you pull the trigger the game detects this and paints the screen black on the next frame.  The photodiode assesses the brightness at this point as a reference for ambient light.  Then on the following frame the game draws one or more white squares on the screen where the duck(s) were.  If the photodiode now registers more light than the reference it captured a frame ago a hit is registered.  There's a bit of faffing about for which of multiple targets are hit, but that's essentially the jist of it, as explained here:

 

 

The only small complication I immediately see to interfacing such a gun to the Intellivision II, SSVA or ECS is that the Inty controller ports don't supply power, and therefore, cannot run the electronics in the gun.  This could be solved by having a battery in an adapter circuit sitting between the gun and the Master Component.  There are plenty of adapter circuits online to interface these guns with other consoles that could provide a jumping off point for such a design.

 

It should also be noted that the NES and Sega guns don't play nicely with modern TVs.  It's not clear to me if this is just down to lag between the console and screen (which might be fixed in new games with a calibration screen and code to expect the lag), or something more fundamental.  Either which way, Inty zapper games would have to be specifically written to take advantage of the gun.  No just plugging in a phaser and blasting away at Space Battle.

 

 

Yes, but I do not think anybody would want or appreciate that sort of flickery screen blanking on their games just to support a zapper controller.  That may have been still a bit of a novelty back in the middle 1980s, but there is a reason it wasn't all that popular.  (It was masked much better back in the older 1970s game machines, wherein the "skeet shooting" games were mostly a white "pong" block on a black background -- but when you have a full colour screen, blanking for two or more frames is not going to win you any friends.)

 

The better implementation, as also described in the video, using the horizontal re-trace timing to detect the position of elements within the same frame would not work on the Intellivision.

 

     -dZ.

[+cmadruga]

I'd love to write a light gun game for the Intellivision platform.

 

 

[mr_me]

8 hours ago, DZ-Jay said:

...

The better implementation, as also described in the video, using the horizontal re-trace timing to detect the position of elements within the same frame would not work on the Intellivision.

What's the reason for this? If a crt scan sensing light gun can send a signal precisely to the Intellivision. Can the Intellivision not measure the time it receives that signal relative to vblank. The Intellivision might be too slow for a pixel precise calculation but it can still be accurate.

[+DZ-Jay]

29 minutes ago, mr_me said:

What's the reason for this? If a crt scan sensing light gun can send a signal precisely to the Intellivision. Can the Intellivision not measure the time it receives that signal relative to vblank. The Intellivision might be too slow for a pixel precise calculation but it can still be accurate.

How can it measure the time since VBLANK?  The only thing I can think of is for it to be completely idle, polling for a signal in a carefully calibrated loop from which it can extrapolate the timing from the known instructions executed in the meantime?

 

We do a similar thing to detect PAL vs. NTSC at runtime, but there is nothing else happening at that point.

 

If there is any variability or non-determinism to the execution path (e.g., any conditional branching based on external feedback, etc.) it won't work.

 

Even if there is a deterministic path, you must interrupt it to poll the signal, which will either severely reduce the granularity of polling, negatively impact game-play mechanics, or both.  (E.g., If you poll once at exactly the same spot in your game loop, your time sensitivity is now reduced to 60Hz; if you poll twice at precise intervals, your sensitivity is 30Hz, but your game is interrupted twice and has not had a chance to do anything that alters its determinism.)

 

Maybe I'm missing something and someone can find a way to do it in a single frame.  That would be great.  However, I do not believe the screen blanking/flashing technique is worth it.

 

Another technique mentioned in the other thread was to have a separate microcontroller in the light-gun itself that takes care of the timing.  What about that?

 

    dZ.

[mr_me]

54 minutes ago, DZ-Jay said:

How can it measure the time since VBLANK?  The only thing I can think of is for it to be completely idle, polling for a signal in a carefully calibrated loop from which it can extrapolate the timing from the known instructions executed in the meantime?

 

We do a similar thing to detect PAL vs. NTSC at runtime, but there is nothing else happening at that point.

 

If there is any variability or non-determinism to the execution path (e.g., any conditional branching based on external feedback, etc.) it won't work.

 

Even if there is a deterministic path, you must interrupt it to poll the signal, which will either severely reduce the granularity of polling, negatively impact game-play mechanics, or both.  (E.g., If you poll once at exactly the same spot in your game loop, your time sensitivity is now reduced to 60Hz; if you poll twice at precise intervals, your sensitivity is 30Hz, but your game is interrupted twice and has not had a chance to do anything that alters its determinism.)

 

Maybe I'm missing something and someone can find a way to do it in a single frame.  That would be great.  However, I do not believe the screen blanking/flashing technique is worth it.

 

Another technique mentioned in the other thread was to have a separate microcontroller in the light-gun itself that takes care of the timing.  What about that?

 

    dZ.

That's right. It only has to that when it receives a signal from the light gun for one frame. It receives a signal from the light gun and the next frame it measures the timing of the next instance of that signal. Is that not possible without negatively impacting gameplay mechanics.

Edited March 24 by mr_me

[8bitwidgets.com]

10 hours ago, cmadruga said:

I'd love to write a light gun game for the Intellivision platform.

hey you were on Casey's show the other day right?   We're definitely going to be looking into this further in the next couple months, so if things still seem feasible we're definitely going to give it a try.  I think we'll start with the phaser first only because of the db9 connection..  eventually the zapper would get the same treatment..  both are still fairly easy to find and run around the same price..   

 

11 hours ago, DZ-Jay said:

I do not think anybody would want or appreciate that sort of flickery screen blanking on their games just to support a zapper controller. 

Anyone who has ever played a light gun game knows the flicker.. and I'm not sure anyone here can really say they don't have some tolerance to a poor gaming experience for the sake of nostalgia

 

1 hour ago, DZ-Jay said:

Another technique mentioned in the other thread was to have a separate microcontroller in the light-gun itself that takes care of the timing.  What about that?

I would only consider that if the normal flashing version wouldn't work.  This is about trying to make use of existing / accessible accessories to avoid having to engineer an entirely new technology.  The more dev / cost goes into that the tougher it is to reach folk.  Especially since on top of that, the tech still requires CRT which I know will already lose enough players.  I have a CRT, but I know many don't, however it's also still not hard to find one..  

 

1 hour ago, mr_me said:

Is that not possible without negatively impacting gameplay mechanics.

I mean from what I'm seeing, if it works like duck hunt and gumshoe, i don't see how it will be a negative impact..  it's how these games worked back then.   i see no reason that new games working with the same limitations will be a turn off.  maybe it won't win over people who were never big fans of light gun games (and bases on how few games were made for it, a lot) which is why making it as true to the real thing and not over-engineering a new solution is key to see this come to life.

[+DZ-Jay]

57 minutes ago, mr_me said:

That's right. It only has to that when it receives a signal from the light gun for one frame. It receives a signal from the light gun and the next frame it measures the timing of the next instance of that signal. Is that not possible without negatively impacting gameplay mechanics.

I am sorry if I am misunderstanding your point, but I do not see how it can be done.  There is no "receives a signal," since there is no way to interrupt the CPU except by the STIC's VBLANK signal.  Therefore, you would need to poll the input port in order to check if a signal arrived.

 

Now, say you detect a signal in one frame, and then a signal in another frame.  Apart from a distance in time of 1/60th of a second (the overall time between frames), what else can you compare against?

 

Say that you want to measure signal #1 from the moment the VBLANK interrupt triggered to the moment the light-gun signal arrived.  Then you want to measure the same for signal #2, and compare their relative distance from their respective interrupts in order to extrapolate the time between the two signals.

 

How do you do that?  You need a source of timing.  You have VBLANK, but that brings your back to 60Hz.  What else is there?

 

Well, there is the timing of each CPU instruction executed -- these are known, of course.  To use that, you need to measure the number of cycles of all instructions executed by your program from the start of VBLANK to the moment you polled the input port.  You need to account for all code, including the code to poll.

 

That is certainly possible -- but just so long as that source of timing is deterministic and constant.  If there are any branches or differences in the execution flow between the two polling instances, you must account for that too.  If there is any variation due to external input (say, anything influenced by a pseudo-random number generator which is itself influenced by user input), how do you account for that?

 

If your game loop is stable, regular, and has a constant execution flow, then this can work -- as long as you account for all the cycles incurred by the code in between intervals.  However, that seems to me to be an overly onerous requirement to impose on a game program.

 

Still, it may not be technically impossible, but it may not be very practical.

 

All that being said, I would be very happy to be proven wrong by an enterprising programmer.

 

    dZ.

[8bitwidgets.com]

4 minutes ago, DZ-Jay said:

If there are any branches or differences in the execution flow between the two polling instances, you must account for that too.

using duck hunt as our gold standard, i'm curious about the branches you're thinking of.  it seems pretty linear.. maybe if two guns were involved?  

 

I read how they could determine more than one target on the screen (though i'm sure 2-3 was pushing it)  From this play through there never are more than 3 targets on screen so that's ever 20th of a second it could poll and check which it is..

 

You can see the stutter the check does as a player shoots so this is a feature not a bug..   par for the course.   

 

 

I never realized how much Gumshoe looks like Walter White.. hmmm 

 

 

[+DZ-Jay]

29 minutes ago, 8bitwidgets.com said:

 

Anyone who has ever played a light gun game knows the flicker.. and I'm not sure anyone here can really say they don't have some tolerance to a poor gaming experience for the sake of nostalgia

 

I am here to say precisely that -- at least for myself.  

That flicker was not a very popular game mechanic, and was rapidly replaced by superior techniques.

 

Moreover, nostalgia brings with it a certain kind of sanitation, a tendency to ignore that negative aspects of the past.  As I have heard it said once, everybody loves to be in the Renaissance Fair, but nobody really wants to live in the Renaissance.  They want the spectacle, the costumes, the roasted turkey leg, the fancy floral language; but nobody wants the diseases, the rotten meat, the sewage flowing through the streets, the smells of human and animal waste, etc.

 

Your mileage may vary, I suppose. *shrug*

 

29 minutes ago, 8bitwidgets.com said:

I would only consider that if the normal flashing version wouldn't work.  This is about trying to make use of existing / accessible accessories to avoid having to engineer an entirely new technology.  The more dev / cost goes into that the tougher it is to reach folk.  Especially since on top of that, the tech still requires CRT which I know will already lose enough players.  I have a CRT, but I know many don't, however it's also still not hard to find one..  

 

From my view, engineering an entirely new technology (already done in the past, so it is not from scratch) may be more productive.

 

Or ... how about we use the technology we already have?  There are other ways we could employ the disc which have yet to be explored ...

 

29 minutes ago, 8bitwidgets.com said:

I mean from what I'm seeing, if it works like duck hunt and gumshoe, i don't see how it will be a negative impact..  it's how these games worked back then.
 

 

They were not very popular games.  They may have been back when there was not much else ...

 

29 minutes ago, 8bitwidgets.com said:

I see no reason that new games working with the same limitations will be a turn off.

 

I do, for the same reason people criticize bad mechanics and clunky controls from the old Mattel games.  We do not try to replicate those, do we?

 

I'm sure the Italian producer of that other Ghostbusters game thought himself that there is no real quantifiable difference between his game and @cmadruga's.  I imagine that more one one person would disagree with that.

 

29 minutes ago, 8bitwidgets.com said:

maybe it won't win over people who were never big fans of light gun games (and bases on how few games were made for it, a lot) which is why making it as true to the real thing and not over-engineering a new solution is key to see this come to life.

Hey, people are free to do what they wish, of course.  I am not telling anybody not to do this.  I am just pointing out some potential pitfalls in the technology and it's market appeal, as I see them.

 

    dZ.

[+DZ-Jay]

14 minutes ago, 8bitwidgets.com said:

using duck hunt as our gold standard, i'm curious about the branches you're thinking of.  it seems pretty linear.. maybe if two guns were involved?  

I do not know if there is any branching.  I am saying that if there was, it would severely complicate matters.  By the way, "branching" in this context refers to any "IF/ELSE" or conditional "GOTO" that alters the execution flow.

 

14 minutes ago, 8bitwidgets.com said:

I read how they could determine more than one target on the screen (though i'm sure 2-3 was pushing it)  From this play through there never are more than 3 targets on screen so that's ever 20th of a second it could poll and check which it is..

 

You can see the stutter the check does as a player shoots so this is a feature not a bug..   par for the course.   

 

 

I never realized how much Gumshoe looks like Walter White.. hmmm 

 

 

The technique you were originally promoting relies on blanking the screen.  You would need to blank the screen once to clear it, then once more for each target you wish to detect.  Three targets means blanking for 4 frames out of 60 in a second -- that is a noticeable gap and detectable by the human eye.  In my opinion, it would be very distracting.

 

The technique @mr_me and I were discussing requires using a secondary source of timing in order to detect the light gun pulse within a single frame.  That secondary source is provided by the Horizontal Blanking signal of the raster in other consoles, but the Intellivision does not provide access to the HBLANK interrupt.  The only other source of timing on the Intellivision that is smaller than 60 Hz is counting the CPU cycles incurred by every instruction executed from the moment the interrupt arrives, to the moment you poll and detect the light gun pulse.  That is not trivial, and requires careful coding and stable execution flow.

 

   dZ.

[8bitwidgets.com]

9 minutes ago, DZ-Jay said:

Your mileage may vary, I suppose. *shrug*

yea i get what you mean, for me it's really more about making something both authentic but actually more importantly not turning a novelty into a monumental task.   

 

10 minutes ago, DZ-Jay said:

There are other ways we could employ the disc which have yet to be explored ...

i'm listening   

 

12 minutes ago, DZ-Jay said:

I am just pointing out some potential pitfalls in the technology and it's market appeal, as I see them.

duly noted   all valid points.  I'm going for the quickest / most affordable / accessible solution first..  gameplay improvements are secondary.  Perhaps if this is successful, and if there is enough interest in it, perhaps we could do something deeper.  

[8bitwidgets.com]

12 minutes ago, DZ-Jay said:

I am saying that if there was, it would severely complicate matters.

oh for sure.  

 

12 minutes ago, DZ-Jay said:

that is a noticeable gap and detectable by the human eye.  In my opinion, it would be very distracting.

yea it was noticeable and still is, but "it was the 80s' man"    We'll start there and then see where it leads for a v2.0 and one ideally that would work on HDMI TVs as well if we get that far. 

 

 

12 minutes ago, DZ-Jay said:

That is not trivial, and requires careful coding and stable execution flow.

Gotcha, yea that is an interesting alternative.  Something to think about in the future (future, future, future..)

 

Edited March 24 by 8bitwidgets.com

[First Spear]

I hope whatever path you choose is successful. Thinking on top of the tech details, it could be important to have something "that just works" for modern TVs. It's probably safe to say that 95% of people with Intellivision-to-TV connection issues think of changing their console to fit their TV rather than finding a TV that works with their console. Even if there was a "brick" in between the console and the gun, people would likely gravitate towards it instead of something sleeker that wouldn't work with their setup. Of course if you don't have large-ish commercial plans none of that matters.  

Thanks for looking into it!

Edited March 24 by First Spear

[Walter Ives]

On 3/19/2024 at 8:10 PM, 8bitwidgets.com said:

Can the Intellivision use a light gun?

While you can get some functionality using the NES approach, here's the preferred way of doing this back-in-the-day.

 

The interior of the display surface of a CRT is coated with a "phosphor." An electron beam is controlled to scan over that phosphor in a "raster" pattern containing 262 horizontal lines. Each horizontal line is scanned in 63.5 microseconds. The beam's intensity is modulated much faster than that: where it is strong it creates a bright spot, where it is weak, it creates a dim spot. A new raster is displayed every 60th of a second. The brightness of any particular spot on the phosphor decays to about half its value every 60th of a second.

 

So, if you stare closely at any particular point on the screen, its brightness will almost instantly jump to a given brightness, then slowly decay to about half that over the next 60th of a second. As long as the spot is reasonably bright, that jump is easily detectable. If you know precisely when the intensity jump of the spot you are looking at occurs, you know where on the screen you are looking.

 

To find the y-position, use an 8-bit counter that is cleared at the start of every active field (at the first SR2 after SR1), clocked at the start of every horizontal line (CBLNK) and stopped when the intensity jump occurs.

 

To find the x-position, use another 8-bit counter that is cleared at the start of every horizontal line (CBLNK), clocked with MCLK and stopped when the intensity jump occurs.

 

Read the resulting x,y position when you get to the end of the field.

 

The Intellivision cartridge port contains all of the signals needed to implement such a circuit: SR1, SR2, CBLNK and MCLK.

 

This was all well understood back-in-the-day. General Instrument even included the required counters in some of its dedicated game chips. See https://web.archive.org/web/20120316160455/http://www.pong-story.com/GIMINI1978.pdf, p. 11. Had Mattel wanted, GI could have provided a very simple custom chip that could have been added to a cartridge for less than a buck. Actually, the cheapest solution would probably have been to just add the required circuitry to a corner of the existing ROM chip design. That would have saved the cost of dealing with an extra package.

 

WJI

 

[Walter Ives]

On 3/19/2024 at 8:10 PM, 8bitwidgets.com said:

So, are we going to be making a light gun for Goose Hunt?!

You can print a 3-D housing for your circuit that matches the styling of Intellivoice or the Playcable adapter. Call it the Intelligun® and hope it sells better than the other two.

 

WJI

 

[8bitwidgets.com]

On 3/24/2024 at 4:59 PM, First Spear said:

"that just works" for modern TVs

This is something we'll look into, but i'm skeptical.  unfortunately it's a CRT world for classic gamers.  I have a CRT for this reason.  I'm open to making something for modern screens, but I'm imagining that's a bridge to far.  The amount of work it would take to make a new gun for such a smaller user base is just not feasible.. far easier to find a CRT for those who actually want to use a light gun.  Mohammad will have to go to the mountain.. heh

 

On 3/24/2024 at 4:59 PM, First Spear said:

Of course if you don't have large-ish commercial plans none of that matters.  

exactly.  I'm just not sure how deep the well is here to engineer a solution.. it feels like it would be right up there with FPGA Intellivision.. totally possible, but the amount of work it would take to make for the size of the audience per unit price would be too high..   I certainly plan to make something that is available for anyone who wants it, but we have to manage the risk of how many people would actually buy it.  

 

As it stands with us, light gun dev is on hold as we're working on other projects that are likely going to be more interesting to a broader audience, but the good news is that if this works out, it likely will be the pathway that gets us to a light gun possibility.  We're definitely going to give it a solid attempt and if we can do it, we will.

 

4 hours ago, Walter Ives said:

The Intellivision cartridge port contains all of the signals needed to implement such a circuit: SR1, SR2, CBLNK and MCLK.

This is the route we're going to explore down the road.  I don't see us trying to do it through the joystick ports because 1. a LOT of people still have wired systems  2. not interested in dealing with batteries and/or external PSU

 

[First Spear]

23 hours ago, 8bitwidgets.com said:

(snip)

 

This is the route we're going to explore down the road.  I don't see us trying to do it through the joystick ports because 1. a LOT of people still have wired systems  2. not interested in dealing with batteries and/or external PSU

 

Since this is all conjecture and you don't have time to work on the hardware, I'll keep going.

 

That might be a good opportunity to use (require) and ECS. All Inty consoles support it and you'd get two additional plug ports as well as 2k more RAM.