After I was ‘dismissed’ from the negotiations, I returned to making the chipsets for the new mainboards.
Nothing spectacular here. The various tiers of chipsets mostly differed in how many UEB lanes they could divert to for example the UEB expansion slots, the OPB headers, or how many memory channels they allowed. In other words, the typical modularity of a modern computer.
I decided to dedicate one UEB controller to low-priority peripherals, like fan headers, coolant pumps, RGB headers for those who needed that, or OPB headers. Yes, it was more than was needed except in the most extreme cases with several dozen OPB ports, a score or two of fans, and a couple of precision-controlled pumps.
But honestly, if a build could not work with the remaining controllers, you have chosen too small a processor. Let’s be honest, anything you need more than 64 UEB controllers for, you also need more processing power than a Hyperion 9 could deliver. And I seriously doubted that there was anything less than a full-blown cyberdeck would ever need more than 70-80.
For desktop systems that is. For servers that was a completely different thing of course. That was the reason why the Chimaera started at 128 controllers and went up to 256. It was of course similar for the Chronos. And even a single Grendel-server would be hard pressed to physically squeeze as many devices into it as it could control.
But back to the mainboard chipset design, the CPU that they were designed to work best with decided on the number of slots available. Or how many mass storage modules it could host.
I have to say, it is almost as complicated to say these things as it was to design the chipsets, so it was way easier than it sounds.
On the other hand, it was significantly harder to create drivers for all the new tech. And not just for Envision, where I decided to support everything from Envision 33 on forward. Linux alone with its who knows how many distributions and forks was a challenge. In the end, I left it with supporting the most important 20 distributions and making the drivers open source.
If somebody absolutely wanted to use a non-standard Linux version, they could adapt the drivers by themselves.
That left drivers for K-DOS, though if our negotiations with Ralcon were successful, the Kawamoto Desktop Operating System would cease its rapid growth in user base. After all, it was another attempt to make a VR-capable operating system similar to Dalgon Mentor.
But where Mentor was mostly directed at students and was optimized for online classes and college work, K-DOS tried to charge into the OS market at large, aiming to displace Envision. It was a clear attempt to take over the market, while Mentor was more of a jury-rigged solution to the problem the Commonwealth had found itself in when it changed the schooling system to virtual reality.
I was pretty sure that Mentor would vanish almost completely as soon as Envision became VR-capable. Not so K-DOS. Kawamoto was actively developing it, and its accompanying office suite. Not that they had managed to get more than a fraction of the functionality of Envision baked in.
Only the VR capabilities even gave them a chance.
Nonetheless, I included drivers for K-DOS and Mentor. Fortunately, Mentor was Unix based similar to Linux, which made it relatively easy for me to create the drivers for that.
Well, to be honest, I mostly did the rough work and left the finicky detail works to Warden. Over the last few months, I came more and more to the realization that I had no clue how I had managed to do my work without her. Oh sure, I could do the work myself. Heck, I had done the work often enough. I even had written not one but two complete OS for my series of cyberdecks.
And yes, I had borrowed liberally from other operating systems. Otherwise, it would have been impossible even for me to get it done. But while it was strongly inspired by Linux, it was not compatible, by design. Every hacker worth their salt was able to penetrate virtually any Linux machine at will. Not that Envision or K-DOS were any better. The point was that in the age of supercomputers and VIs it took only a few weeks before a new version of an os was deconstructed.
But as far as I knew, my OS was rare enough that nobody bothered to put in the effort. That was the reason why so many different firewalls and antivirus programs existed. The corporations were hoping that if the market was fragmented enough at least some of the security programs would remain unbroken.
That was also the reason why I had two versions of my OS per generation of boards. One that I sold, and one that was only for me. I had done my best to move the vulnerabilities around. I was pretty sure that I was only marginally successful in that endeavor, but it was better than giving anybody who might be antagonistic in regards to me access to my personal OS.
But without Warden, I would have to spend the next two or three weeks working on the drivers. Despite the insane compression I had. I was not sure how much computing power Warden had by now, but I was pretty sure it was enormous. My guess was that she had at least a dozen fully stocked Grendels up and running. That allowed her to be exceptionally fast in modifying code. Not in devising it, but if I gave her an objective, she usually finished within a few minutes.
As it was, it took me longer to conceptualize the required functionality than it took for Warden to complete the finished drivers.
Sure, it took her some time to test the new drivers to exhaustion, but that was mostly because she had to build the hardware before she could test the drivers. It would take her a few days to have everything tested.
I was relatively sure that the early versions of the drivers were at least moderately usable by that time though. Thus I decided to build a few of the high-end systems, just to place them in the hands of some of the people I worked with.
Yes, Jacky first and foremost, but also Ben and Michael. After a bit of contemplation, I decided to add a machine of Doc Schaeffer to the list. Four Hyperion 12 1990G and four Theia 90 were baking, alongside all the boards and the memory, when I remembered that the off-the-shelf holostorage-systems would just limit the whole system.
I had bypassed that bottleneck in Glory by adding enough additional volatile memory to mirror the most important content into a RAM disc. Unfortunately, there was no place for that in the way that computers were constructed.
Oh sure, I could modify the mainboard to include an additional row of memory slots, but that was a rather clunky solution. No, what was needed was a redesign of the holostorage itself.
The basic principle of it was already as good as it was likely to get. A matrix of dark and light sections was written with a group of UV lasers into an otherwise clear medium. The matrix was only illuminated if another UV laser with a very specific frequency hit it at a specific angle. In a standard quadratic crystal with a width of close to 2cm that translated into roughly 2.6TB per matrix.
Usually, a crystal stacked a few thousand matrices in this fashion.
The really neat trick was that by changing the angle or frequency of the illuminating laser it was possible to write several matrices at the exact same position without interfering with each other.
The incredible speed these systems possessed came from them reading or writing a complete matrix in one go. No sequential read, just the whole matrix at once.
That was also what had made the HNVMH, or Holographic Non-Volatile Memory Host necessary. Before holostorage, all storage options were working sequentially and were best served with a serial bus. The HNVMH on the other hand was massively parallel to facilitate this enormous bandwidth. That was only exacerbated when manufacturers began to include several arrays of writing and reading laser working in parallel.
And as with so many other things, this technology had hit a hard wall of its possible performance quite some time ago. Not that this was in any way a problem, as only the most powerful supercomputers could come even close to hitting the limit of what holostorage could provide.
Until now that is. The HNVMH was, fortunately vastly over-dimensioned to what was needed. It was ‘future-proved’ at a time when the physical limits of the holostorage technology were barely understood, and it was easier to just keep it as is than to change the standard. But that left me with the physical limits themselves.
I could switch over to x-ray lasers or gamma-ray lasers. If I was completely inconsiderate of the continued wellbeing or survival of the users that is. This step has been considered, long ago. But it was deemed impossible to shield the harmful radiation. So, no joy here.
The lesser informed being might get the idea to simply make the crystal bigger. A bigger crystal equals more space equals a bigger base matrix, which equals higher capacity and bandwidth. A completely logical conclusion. Unfortunately, the material that made up the crystals became increasingly opaque and allowed fewer angle and frequency combinations the thicker it became. It had been determined long ago that the most efficient width of the crystal was 19.387mm.
That left me with a variation of what I had done for Glory. But instead of building the RAM into the computer, I decided to build it into the storage system. Something of a super cache, combined with a RAM disc. Naturally, even with a multi-level ram in the 414pm design, it was not possible to mirror the whole crystal into the ram.
Unauthorized reproduction: this story has been taken without approval. Report sightings.
No, I decided to include 12 PB of semi-static RAM-disc and 16 PB of superfast cache into the design. Future programmers would have to designate parts of their programs as important for fast execution, to make the most use of the available ram, it just couldn’t be helped.
Don’t get me wrong, I did not expect the average application, even the average operating system to burst out of the ram I had included anytime soon. It was a different matter for a cyberdeck, and would likely be the defining factor for speed in the future.
The point here was that the OS of a board had an insane amount of functionality baked in, which needed to be ready at any given time. Add in that the OS and the utilities of a board were normally optimized for execution speed, instead of minimalized size, well you get the picture.
Glory’s OS and utilities had a size of nearly 13 EB or more than three orders of magnitude bigger than what I was building into the new design of holostorage. That meant that she also had an absolutely ludicrous 16 EB of RAM-disc. That was only possible through the Q-link, and even then it required extreme integration of the system. And Glory still had nearly twice the volume of Precious.
Sure, she was smaller than the average desktop PC, but that was not saying much, considering that those were usually not quite considered mobile.
But I digress. I believed that at least for the next few years, 28 PB of ram inside the individual holostorage systems would be good enough. Honestly, even very resource-intensive games rarely even approached this size. Envision did not even have a size of 2 PB, though knowing the industry, I was relatively sure that as soon as the next generation of hardware became available, the size of it would… explode.
If the code monkeys and project managers of Ralcon could refrain from adding every single pet project and prestige feature that they could not squeeze in by now for the sake of performance I would eat replicator food for a week.
But what it boiled down to was another part of the computer I was designing that needed to be built by the NADA. It was already clear that three NADA were not enough.
Heck, I would probably have to try to design something of an industrial-sized NADA in the foreseeable future, just for Enki to be able to produce many of the toys I was developing.
But that was for the future. For now, I had my three NADA busy making the parts for the new computers. At least the parts that needed to be made by the NADA.
Other things, like the holographic emitter, the crystals for the holostorage, and even all the cases, I tasked the industrial fabber with. No need to clog up the still relatively slow NADAs with something every semi-decent workshop could do.
It would still take a couple of days for all the parts to be ready. As it was now roughly time for dinner, I surfaced and stretched for a bit.
Darren and Ryan were cooking that day, and they had become quite good by now. It helped that I could provide spices on demand already. Technically, I could also provide as much meat as was wanted, but for now, I preferred organically grown meat.
We were having beef scallopini ai funghi, fortunately for me without any of the alcohol of the white wine left in the sauce.
I had to say, it was astonishing how good at cooking most of the others had become. Yes, we still did let neither Mark nor Christine use real ingredients, as they had proven to not being able to cook in VR, but all in all, the harvest of adequate chefs was exemplary.
At first, we had the usual, shallow conversation, but when the flurry of greetings and vacuous remarks about the weather ebbed, I decided to change the topic.
“By the way, Jacky, your new computer will be ready the day after tomorrow.”
It took her a moment to answer:
“My… new computer? Why do I get a new computer? And when did you decide that?”
Now it was my turn to be confused.
“I thought that was clear when we talked about the new VR mode for Envision.”
Instead of Jacky answering, it was Natalie who exclaimed: “Wait, Envision has a VR mode? Why the fuck do I not know that?” Followed by most of the others with similar sentiments.
When the general outbreak had calmed down a bit, I was able to answer:
“It doesn’t. Not officially at least. Yet. The VR mode for Envision is what we want to offer Ralcon for their, well, let’s say neutrality. Jacky is now testing it for me.”
The woman in question smiled and nodded.
“Yeah, about that. It works fine, but sometimes it gets… I don’t know, kinda slow?”
I shrugged.
“Not surprising. What I gave you to test it with is literally the smallest computer that exists. Basically just enough to do office work on and that’s it. But otherwise, everything is fine?”
“Yes, much better than the old keyboard and mouse control. But that… I don’t know, it is always so bothersome when it slows to a crawl.”
“Hm, I might have to look into the resource manager then. Maybe have a watchdog look into what you have up and running. If you only use Envision Office the garbage can should not slow down that much.”
As an answer, Jacky giggled nervously.
“Uh, only Office? Uhm… that… might be the problem. I usually have around a dozen browser tabs, a media player, several instances of the office tools, and a few other stuff up and running.”
“Well, there you have it. Even without VR that would tax the garbage can you have now.”
“But it is no fun just to do Office work.”
“Well, that is what the new computer is for. That one is anything but a garbage can. It has many of the components I have in Glory. Not quite on the level of Glory, as it is more modular and that always costs performance, but the top of the line anyway. You literally can’t buy anything similar right now.”
“Oh cool. So, will that thing make the VR run a bit faster? I would love to be done with my work a bit faster so I have a bit more time over.”
I shook my head.
“Not with the diadem. That is limited to 4:1. To get faster, you need a jack. By the way, why haven’t you asked for one yet?”
She shrugged but made a disappointed face.
“Those things are kinda… I don’t know, expensive, you know?”
I frowned. Expensive? Even with just the monthly stipend, I paid all of them, they would have enough to get an ultra-bandwidth jack with the cranial board. Easily. They had virtually no costs of living. I owned the building, bought the ingredients, and provided energy and water.
“Expensive? The ultra-jack with the cranial board costs only $1300 to make.”
“That may be, but if you remember, you told us that it would be expensive to have the cyberware adapted.”
I did? When did I tell them… oh… right.
“I think I remember. Do you remember that that was at the very beginning of us being here? At that time, I did not know you from Adam. So yes, I told a bunch of strangers that I could provide some services for a fee. But we are no longer strangers, are we?
Also, that was before we began to make cyberware literally on an industrial scale. Before I had the industrial fabber even. Those prices I told you then were when I expected to get the cyberware from the black market, and when I expected to have to spend several weeks per implant to adapt them.”
Christine tilted her head.
“Does that mean that we can get a jack for cheap?”
“Yes, of course. Well, most of you. Sadly, Darren and Kate can’t use off-the-shelf jacks.”
That made Ryan flare up: “Now, wait a minute here. Why Darren and Kate? I haven’t pegged you as a racist.”
Darren put a hand on his lover's shoulder.
“Calm down. Remember the problems Kate and I had with the diadems? Even now they don’t work right.”
“Exactly that.” I agreed with him. “The two of them simply have a different neural structure. I can make them jacks. Probably. But those won’t be mass-produced. I will have to basically handcraft the implants for them. And unlike with you norms and me, there is some risk involved.
I will have to clone their brains to test the cyberware, and even then, it is not completely without risk, as I categorically refuse to have the clone brains wake up and suffer.”
Darren rubbed his chin, while he looked at me.
“Hm, how much risk are we talking about?”
I shrugged.
“Not much. Maybe 2%. Maybe 4%. Of it not working right that is. The risk of damage to the brain is significantly lower. Under 1%. But it is still there.”
“I will think about it. It would also depend on what it would cost.”
Kate just nodded when Darren voiced his opinion on the matter.
Christine on the other hand snapped her fingers.
“Back to us normies, you say we can have a jack at cost?”
“Sure. The complete package, ultra-jack, and cranial board for $1300. It takes roughly 30 minutes for the process and another hour or so before the anesthetic wears off. If you want a lower jack… well I won’t stop you but I will be sorely disappointed. Oh, and of course, we have to have the implants delivered first. You have to make it out for yourself in what order you go into the surgeon.”
The rest of the discussion consisted mostly of them working out who was first and so on. I meanwhile sent an order for the jacks for them to Warden, who prepared the delivery from Enki.
Their arguing about the order was still in full swing when Michael came into the cafeteria.
“Ah, you’ve already eaten? Darn.”
That prompted Darren to jump up and put another portion for Michael together.
Not that the others cared all that much about Michaels's presence. I on the other hand was more concerned about him.
“So, how has it gone?”
“Good. We will get $14 for each bundle of neuronect.”
“$14? I thought we decided on $12.”
“Ah yes, as I’ve learned, the producers of neuronect have just jacked up the price to $16 per bundle. For preferred customers like Enertech. Ms. Manning offered with not wanting to go higher than $15. I countered with $14 and we all were happy.”
“That is nice. And the rest was alright too?”
“Yup, after I kicked you out it was only about the nuts and bolts.”
“Do you have an idea what has gone wrong in the beginning?”
He shrugged.
“I can only speculate, but as I see it, Barton and Howard are fuck-ups and they fucked up by the numbers.”
“What do you mean they are… uh… feck-ups?”
He laughed softly when I stumbled over the swear word.
“As far as I can tell, those two are only still working for Enertech because they have connections. They have too valuable contacts to be simply thrown out but are too incompetent and insubordinate to give them real work.
So they will be used for no-effort negotiations. Where they only have to walk in, find out what the other side is offering, and then agree.”
I could not help and frown.
“We are… a no-effort negotiation?”
“For Enertech? You bet. View it from their perspective. We have nothing that is attacking their core business, and the only things where we do impact them is where we will most likely increase the whole market substantially, including their market share. So they have no problem with us.
On the other hand, what we want from them is just to let us be. That they don’t attack us. They had no intention to do that anyway, so for them, anything we would give them to do what they would have done anyway was just gravy.
In other words, from their perspective, the whole negotiation was just to find out what we would gift them.”
“But that doesn’t explain the hostility of Barton and Howard.”
“As I said, those two fucked up. As I understand it, those two have an overinflated opinion of their own importance. They ignored the briefing and got the glorious idea of browbeating us into submission.
I think the powers that be at Enertech have thrown Powell into there to shore up the negotiations for them, but Barton and Howard are too much in a rut. And have dragged him down. No, they had the orders to get what we were willing to give them, rubber stamp the agreement, and be done with it.
And they fucked that up royally.”
I shook my head.
“And the way you are describing it, they are still too well connected to get thrown out, right?”
“Yup, but that is not our problem. We have our agreement with them. And not even just neutrality, but Enertech is actually a tiny bit on our side.
Of course, we also can expect stiff opposition from Cybertronic and Deltatech.”
“Those are B-level corporations, right? We can live with them being against us.”
“Yup, B-levels, and the two main providers of neuronect on the American continents. That is virtually all they produce, so they will be gone quickly.”
“Hm, ok. We always knew that we would have to stomp on some toes to go forward. And those two would have been victims either way.”
The others had, by now, finally finished deciding on the order they would crawl into the auto-surgeon.