• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Editorial Moore's Law - Is it Really Dead ?

AleksandarK

News Editor
Staff member
Joined
Aug 19, 2017
Messages
2,190 (0.91/day)
"Moore's Law" is a term coined in 1965 by Gordon Moore, who presented a paper which predicts that semiconductor scaling will allow integrated circuits to feature twice as many transistors present per same area as opposed to a chip manufactured two years ago. That means we could get same performance at half the power than the previous chip, or double the performance at same power/price in only two years time. Today we'll investigate if Moore's Law stayed true to its cause over the years and how much longer can it keep going.





For a period longer than five decades the rule of Moore's Law has worked, making the above graph look almost linear if a line would be drawn. What draws the line away from its linearity are occasional hiccups in manufacturing that the industry had to deal with and did overcome in the end. Through history, Moore's Law has been declared and predicted to be dead many times, as the industry approached sub-micron sizes. Starting from 1 µm, many people became skeptical in the law's viability and how well would it hold up, but time has proved these people wrong and we are now already using products that are based on 7 nm node.

The challenge
What really is trying to stop the law are three things: lithography, impurities and economies of scale.

First problem to address are impurities found at the atom level. Silicon manufacturing foundries are melting quartz sand to form a big crystal structure that is cut up into wafers. When sand is melted, impurities are inevitable. That is due to the fact that heating silicon makes it easily react with the halogens (fluorine, chlorine, bromine, and iodine) to form halides. Those halides are removed by using more chemicals that dissolve and remove them so a big crystal can be extracted for making of wafers. "What's the problem caused by removal of these impurities?", you might ask. The smaller the size of an individual transistor, chances are better that even a small atom of impurities, left over from previous cleaning, will be present and render a transistor unusable.

The second problem to address is lithography. In order to make the transistors, you have to "print" them on a silicon wafer. That is done by shining light through a mold called mask, which penetrates the silicon and etches the pattern into a wafer. Naturally, as you decrease the size of your transistor, you have to make your mask smaller. The smaller the mask, the harder it is to make it (you start to notice a certain pattern here with size decreases). So to solve this problem, the semiconductor industry has made a multi-patterning technique which takes several turns into the etching process in order to boost the efficiency of the mask application. However, often times this approach is not enough and UV light is having trouble printing designs to silicon. That's how Extreme UV or EUV lithography was born. It uses a stronger light source with a shorter wavelength to better etch the design and thus reduce the errors the printing might encounter. What the problem really is here, is not the light that passes through the mask, it is the mask itself. The mask is the critical design element as it transfers your design to silicon. If you can not make precise and small masks, you can not get a working chip. Thus, the making of the mask is another critical step that makes Moore's law hard.


The third and final problem is economics of scale. That is where the less known Moore's second law comes in, which predicts that the cost of setting up a new manufacturing facility is also getting two times more expensive every two years. Today in order to build a new fab, companies are spending billions of dollars. Intel has invested more than $12B in its fab number 42 in Arizona, supposed to manufacture 7 nm chips some day. In addition to the huge capital needed to just open a new fab, companies need to develop its own semiconductor node process. To put things into perspective, industry speculations are that starting from 5 nm and below, over five billion dollars are needed for R&D alone. This is the reason why only three foundries are left manufacturing 7 nm and below - Samsung, Intel, and TSMC.

The way forward

For all the companies that have capital to invest in new manufacturing facilities and equipment, there are quite a few options to choose from in order to keep the law going. Adding new materials, making new kinds of transistors, and stepping into the 3rd dimension.

Knowingly introducing small amounts of other materials into with silicon ("doping") can be a double-edged sword. A new material can boost properties of the transistor, but prove incredibly difficult to manufacture. That is what Intel's experience with cobalt was. They added it to the 10 nm node to decrease resistance in extremely small wires connecting the transistors. Copper is usually used for these wires, but as it is packed into smaller wires, it tends to become more resistive, so cobalt is added beacuse at the same size, Intel found that it is has half the resistance than similar wires made from copper. This addition proved to be useful, but quite difficult to manufacture and it yielded poorly, leading to delays in the new process. Despite the delays, its addition solved a big problem engineers faced, showing the potential of integrating new materials to boost performance. If you remenber, aluminium was used for some time before the industry switched to copper for better performance charactertistics. That transition didn't go smoothly either, but it turned out quite well after a bit of time.

New kinds of transistors are also an option. For a while the industry used standard, planar CMOS FET as a base transistor, which worked fine until we couldn't control current that passes through the transistor, making random switches that often resulted in errors. Just recently, new design called FinFET replaced planar FET, where the fin was raised and the gate started surrounding the source to better control if the transistor will switch or not. In the image below you can see the difference that appeared with introduction of FinFET, allowing manufacturers to make smaller transistors and most importantly control them.



The latest approach to making transistor is "Gate All Around FET" or GAAFET in short. Its design wraps the whole source with a gate, to prevent any possible switches without intent. Planned for use in 5 nm and below, GAAFET is a technology we will see very soon. It will allow even smaller transistor designs with easier manipulation of turning on/off.



And last but not the least is the 3rd dimension. When we step below 1 nm and start measuring node size in picometers, many forces will prevent transistors from getting smaller. You can go small but you can't break the rules of physics. Quantum tunneling is more present at smaller distances, so at one point we can not go smaller in design without having the transistor make the switch at random times. So when we hit the limits, there is still one place where transistors can be put and that is the vertical axis. If we stack transistors on top of each other, we can automatically double, triple or even quadruple the number of transistors per square millimeter, making the potential of this approach quite big. We already use this technology on HBM memory, and it is about to transfer to logic as well. TSMC also makes Wafer-on-Wafer packages which allow for stacking wafers on top of each other, so it isn't impossible to go 3D and pack more performance in same area, but heat, especially heat density can become a problem.

Summing it all up

My personal opinion is that Moore's Law will not end soon. Not this year, not the one following, nor 2025, when Gordon Moore himself predicts the law to end. It isn't going to be an easy fight for silicon manufacturers, but new technologies are already being worked on and some of them will be deployed very soon, like GAAFET, cobalt, and Wafer-on-Wafer which will allow for additional performance improvements. Chip packaging is getting very good with the appearance of chiplets, making system design seem more like a LEGO building, rather than chip design, were you can pack many different chips next to each other, without the need for a PCB between them.

The thing about Moore's law getting hard to follow is that manufacturers have to get creative if they want to compete and make more money, and that gives this challenge a certain beauty which is only visible if we look at the bigger picture and realize that the best and most interesting solutions are to follow the seemingly boring years of easy performance gain.

View at TechPowerUp Main Site
 
Joined
Aug 13, 2010
Messages
5,379 (1.08/day)
Great article.

While great engineers in this industry also claim that this law basically keeps going, internet comments and videos by people who have no idea how silicon engineering works claim otherwise, and sometimes i'm finding myself really confused not knowing who to believe.
 
Joined
Jan 24, 2011
Messages
272 (0.06/day)
Processor AMD Ryzen 5900X
Motherboard MSI MAG X570 Tomahawk
Cooling Dual custom loops
Memory 4x8GB G.SKILL Trident Z Neo 3200C14 B-Die
Video Card(s) AMD Radeon RX 6800XT Reference
Storage ADATA SX8200 480GB, Inland Premium 2TB, various HDDs
Display(s) MSI MAG341CQ
Case Meshify 2 XL
Audio Device(s) Schiit Fulla 3
Power Supply Super Flower Leadex Titanium SE 1000W
Mouse Glorious Model D
Keyboard Drop CTRL, lubed and filmed Halo Trues
This article kind of misses the "density" part of Moore's law. Transistor budgets have continued to increase, but transistor density has been diverging from the 2x2 rule for quite a while now.

Just as a bit of an illustration (data courtesy of Wikichip):
132779


This assumes Intel's 45 nm density in 2006 as an arbitrary starting point, but the results can only get worse if we look back longer than this.
 
Last edited:
Joined
Nov 13, 2007
Messages
10,209 (1.71/day)
Location
Austin Texas
Processor 13700KF Undervolted @ 5.6/ 5.5, 4.8Ghz Ring 200W PL1
Motherboard MSI 690-I PRO
Cooling Thermalright Peerless Assassin 120 w/ Arctic P12 Fans
Memory 48 GB DDR5 7600 MHZ CL36
Video Card(s) RTX 4090 FE
Storage 2x 2TB WDC SN850, 1TB Samsung 960 prr
Display(s) Alienware 32" 4k 240hz OLED
Case SLIGER S620
Audio Device(s) Yes
Power Supply Corsair SF750
Mouse Xlite V2
Keyboard RoyalAxe
Software Windows 11
Benchmark Scores They're pretty good, nothing crazy.
HPCwire: Is Jim Keller concerned about Moore’s law?
Not even a little. Moore’s Law is not dead, and we will elaborate on our plan to continue it over time. My observation is that the end has always been near, but there has always been and always will be new things we can do to keep pushing that end out. The real challenge is, what happens as we continue to scale technology? What new architecture do we need and how do the problems change? That’s the interesting place.


 
Joined
Jun 16, 2016
Messages
409 (0.14/day)
System Name Baxter
Processor Intel i7-5775C @ 4.2 GHz 1.35 V
Motherboard ASRock Z97-E ITX/AC
Cooling Scythe Big Shuriken 3 with Noctua NF-A12 fan
Memory 16 GB 2400 MHz CL11 HyperX Savage DDR3
Video Card(s) EVGA RTX 2070 Super Black @ 1950 MHz
Storage 1 TB Sabrent Rocket 2242 NVMe SSD (boot), 500 GB Samsung 850 EVO, and 4TB Toshiba X300 7200 RPM HDD
Display(s) Vizio P65-F1 4KTV (4k60 with HDR or 1080p120)
Case Raijintek Ophion
Audio Device(s) HDMI PCM 5.1, Vizio 5.1 surround sound
Power Supply Corsair SF600 Platinum 600 W SFX PSU
Mouse Logitech MX Master 2S
Keyboard Logitech G613 and Microsoft Media Keyboard
Transistor density improvements have definitely slowed down, and are now facing a lot of technical challenges in the physical possibility of shrinking transistors at the atomic level, but AMD's work with multi-chip modules might fulfill the general theme of Moore's Law which is performance increase over time. We may get bigger chips, but we still have a path for using more transistors as a single processor, at least in the short term.
 

ppn

Joined
Aug 18, 2015
Messages
1,231 (0.39/day)
Polaris Jun 29th, 2016 Transistors10,300 million
Die Size251 mm²
41,0 Mtr/mm²

Navi Jul 7th, 2019 Transistors5,700 million
Die Size232 mm²
24,5 Mtr/mm²

3 years = 1.66 performance and density improvement. dead.
 
Joined
Mar 18, 2008
Messages
5,717 (0.98/day)
System Name Virtual Reality / Bioinformatics
Processor Undead CPU
Motherboard Undead TUF X99
Cooling Noctua NH-D15
Memory GSkill 128GB DDR4-3000
Video Card(s) EVGA RTX 3090 FTW3 Ultra
Storage Samsung 960 Pro 1TB + 860 EVO 2TB + WD Black 5TB
Display(s) 32'' 4K Dell
Case Fractal Design R5
Audio Device(s) BOSE 2.0
Power Supply Seasonic 850watt
Mouse Logitech Master MX
Keyboard Corsair K70 Cherry MX Blue
VR HMD HTC Vive + Oculus Quest 2
Software Windows 10 P
Polaris Jun 29th, 2016 Transistors10,300 million
Die Size251 mm²
41,0 Mtr/mm²

Navi Jul 7th, 2019 Transistors5,700 million
Die Size232 mm²
24,5 Mtr/mm²

3 years = 1.66 performance and density improvement. dead.


Well based on your words, it is dead for AMD GPU for sure. But we all know their GPU department has been struggling a lot for the past several years.
 
Joined
Sep 24, 2019
Messages
64 (0.04/day)
This article kind of misses the "density" part of Moore's law. Transistor budgets have continued to increase, but transistor density has been diverging from the 2x2 rule for quite a while now.

Just as a bit of an illustration (data courtesy of Wikichip):
View attachment 132779

This assumes Intel's 45 nm density in 2006 as an arbitrary starting point, but the results can only get worse if we look back longer than this.

SRAM cells do not see the same sort of geometric scaling (in size) with lithographic process advancements as logic gates. The same goes for I/O blocks. This chart is not a good counter-point to the idea that Moore's law is in fact dead. Logic cell density is still scaling very well. The old 18-24 month rule has turned into 24-36, but Moore's "Law" is still alive until we run into the fundamental limits of the silicon atom as a semiconducting material likely somewhere in the < 3nm range.

Polaris Jun 29th, 2016 Transistors10,300 million
Die Size251 mm²
41,0 Mtr/mm²

Navi Jul 7th, 2019 Transistors5,700 million
Die Size232 mm²
24,5 Mtr/mm²

3 years = 1.66 performance and density improvement. dead.

You have the transistor count flipped.
The scaling is 80.7% (not 1.66x or 66%)
You can't compare process nodes across different foundries. Polaris was produced by Glofo, Navi is produced by TSMC. The "nm" name given to an individual process node is largely a marketing term meant to indicate progress, not representative of a linear shrink compared to a prior node.
 
Joined
Sep 17, 2014
Messages
20,776 (5.97/day)
Location
The Washing Machine
Processor i7 8700k 4.6Ghz @ 1.24V
Motherboard AsRock Fatal1ty K6 Z370
Cooling beQuiet! Dark Rock Pro 3
Memory 16GB Corsair Vengeance LPX 3200/C16
Video Card(s) ASRock RX7900XT Phantom Gaming
Storage Samsung 850 EVO 1TB + Samsung 830 256GB + Crucial BX100 250GB + Toshiba 1TB HDD
Display(s) Gigabyte G34QWC (3440x1440)
Case Fractal Design Define R5
Audio Device(s) Harman Kardon AVR137 + 2.1
Power Supply EVGA Supernova G2 750W
Mouse XTRFY M42
Keyboard Lenovo Thinkpad Trackpoint II
Software W10 x64
and that gives this challenge a certain beauty which is only visible if we look at the bigger picture and realize that the best and most interesting solutions are to follow the seemingly boring years of easy performance gain.

Well spoken sir!
 
Joined
Jun 18, 2015
Messages
563 (0.18/day)
Polaris Jun 29th, 2016 Transistors10,300 million
Die Size251 mm²
41,0 Mtr/mm²

Navi Jul 7th, 2019 Transistors5,700 million
Die Size232 mm²
24,5 Mtr/mm²

3 years = 1.66 performance and density improvement. dead.
Unsuccessful troll, u should've said:
"How about this,
Bulldozer had 1.3x more transistors than phenom at the same "total" performance (8vs6) but were actually %20 slower core to core."
But no, despite both of these examples, exceptions don't break the rules
 
Joined
Mar 20, 2019
Messages
412 (0.22/day)
Location
Australia
System Name Ryzen
Processor AMD Ryzen 7 5700X
Motherboard Asus TUF Gaming B550-Plus (Wi-Fi)
Cooling Cryorig H7
Memory Kingston Fury Beast DDR4 3200MHz 2x8GB + 2x16GB
Video Card(s) Sapphire NITRO+ Radeon RX 6700 XT GAMING OC
Storage WD_Black SN850 500GB NVMe SSD + Adata XPG SX8200 Pro 512GB NVMe SSD
Display(s) Gigabyte G27QC
Case NZXT H510 Flow
Audio Device(s) SteelSeries Arctis Prime
Power Supply Corsair RM650x Gold 650W
Mouse Logitech G502 Hero
Keyboard HyperX Alloy FPS Cherry MX Blue
Software Windows 11 Pro
Great article, thanks.
 
Joined
Aug 30, 2006
Messages
7,195 (1.12/day)
System Name ICE-QUAD // ICE-CRUNCH
Processor Q6600 // 2x Xeon 5472
Memory 2GB DDR // 8GB FB-DIMM
Video Card(s) HD3850-AGP // FireGL 3400
Display(s) 2 x Samsung 204Ts = 3200x1200
Audio Device(s) Audigy 2
Software Windows Server 2003 R2 as a Workstation now migrated to W10 with regrets.
Moores law and transistor density vs. chip complexity and programming/compiler/microcode inefficiencies. I started out writing 6502 machine code. Lean. No wasted code, machine cycles, or memory. In 1980 those were scarce resources. We can develop code 100x faster today with modern languages and compilers, but they are in no way efficient compared to hard core machine code. I wonder, on average, exactly how inefficient modern compiled code is running in a typical OS, compared to effecient machine code. (Lost speed, wasted watts).
 
Joined
Jan 24, 2011
Messages
272 (0.06/day)
Processor AMD Ryzen 5900X
Motherboard MSI MAG X570 Tomahawk
Cooling Dual custom loops
Memory 4x8GB G.SKILL Trident Z Neo 3200C14 B-Die
Video Card(s) AMD Radeon RX 6800XT Reference
Storage ADATA SX8200 480GB, Inland Premium 2TB, various HDDs
Display(s) MSI MAG341CQ
Case Meshify 2 XL
Audio Device(s) Schiit Fulla 3
Power Supply Super Flower Leadex Titanium SE 1000W
Mouse Glorious Model D
Keyboard Drop CTRL, lubed and filmed Halo Trues
SRAM cells do not see the same sort of geometric scaling (in size) with lithographic process advancements as logic gates. The same goes for I/O blocks. This chart is not a good counter-point to the idea that Moore's law is in fact dead. Logic cell density is still scaling very well.

Fair enough. I only have CGP for most of the nodes I used above though, so I don't think I can calculate the logic gate size. It's notable, though, that the SRAM bitcell size does still exhibit geometric scaling, just at a lower rate than a doubling every two years.
 

Chloefile

S.T.A.R.S.
Joined
Dec 16, 2012
Messages
10,877 (2.64/day)
Location
Finland
System Name 4K-gaming
Processor AMD Ryzen 7 5800X
Motherboard Gigabyte B550M Aorus Elite
Cooling Custom loop (CPU+GPU, 240 & 120 rads)
Memory 32GB Kingston HyperX Fury @ DDR4-3466
Video Card(s) PowerColor RX 6700 XT Fighter
Storage ~4TB SSD + 6TB HDD
Display(s) Acer 27" 4K120 IPS + Lenovo 32" 4K60 IPS
Case Fractal Design Define Mini C
Audio Device(s) Asus TUF H3 Wireless
Power Supply EVGA Supernova G2 750W
Mouse Logitech MX518 Legendary
Keyboard Roccat Vulcan 121 AIMO
VR HMD Oculus Rift CV1
Software Windows 11 Pro
Benchmark Scores It runs Crysis remastered at 4K
Thanks to Intel's 14nm++++++++++++ and MOAR CORES for 4 years old Skylake, I believe it is.
 
Joined
Jul 5, 2013
Messages
25,559 (6.52/day)
Great article.

While great engineers in this industry also claim that this law basically keeps going, internet comments and videos by people who have no idea how silicon engineering works claim otherwise, and sometimes i'm finding myself really confused not knowing who to believe.
To be fair, the number of transistors is not the exclusive metric measured in Moore's law. I agree with the article that it will not end in the next few years, but has already slowed down..

I wonder, on average, exactly how inefficient modern compiled code is running in a typical OS, compared to efficient machine code. (Lost speed, wasted watts).
It's not as bad as one would think, comparatively. But it's not going to run as well as the ASM you worked with.
 
Joined
Aug 6, 2017
Messages
7,412 (3.05/day)
Location
Poland
System Name Purple rain
Processor 10.5 thousand 4.2G 1.1v
Motherboard Zee 490 Aorus Elite
Cooling Noctua D15S
Memory 16GB 4133 CL16-16-16-31 Viper Steel
Video Card(s) RTX 2070 Super Gaming X Trio
Storage SU900 128,8200Pro 1TB,850 Pro 512+256+256,860 Evo 500,XPG950 480, Skyhawk 2TB
Display(s) Acer XB241YU+Dell S2716DG
Case P600S Silent w. Alpenfohn wing boost 3 ARGBT+ fans
Audio Device(s) K612 Pro w. FiiO E10k DAC,W830BT wireless
Power Supply Superflower Leadex Gold 850W
Mouse G903 lightspeed+powerplay,G403 wireless + Steelseries DeX + Roccat rest
Keyboard HyperX Alloy SilverSpeed (w.HyperX wrist rest),Razer Deathstalker
Software Windows 10
Benchmark Scores A LOT
D

Deleted member 185158

Guest
Looking at averages, 2 years give or take a few months seems somewhat accurate..

This PDF shows transistor count and size from 1971 to 2013.
Somebody with time could do the map and fill in 2013 - 2019 with a little research.


I'll give Mr. G Moore an easy 6-12 month leeway personally. close enough for me. Gotta have some margin of error. Economics alone could hold back any technology from moving forwards.
 
Joined
Oct 21, 2006
Messages
621 (0.10/day)
Location
Oak Ridge, TN
System Name BorgX79
Processor i7-3930k 6/12cores@4.4GHz
Motherboard Sabertoothx79
Cooling Capitan 360
Memory Muhskin DDR3-1866
Video Card(s) Sapphire R480 8GB
Storage Chronos SSD
Display(s) 3x VW266H
Case Ching Mien 600
Audio Device(s) Realtek
Power Supply Cooler Master 1000W Silent Pro
Mouse Logitech G900
Keyboard Rosewill RK-1000
Software Win7x64
Performance hasn't doubled in over 10 years; the performance increment has been ~15-20% between generations, if that.

The current single thread performance is only 2.5x between my i7-920 system and the top of the line systems now.
I bought those in 2009.

my 3930k is only a 1.5x difference.

Those are at stock clocks; my processors are seeing much better performance than the benchmarks I'm quoting.

Video cards have seen a lot of growth; In 2009 I was using a HD4870; there were great increases thru the 5870, 7970, Crossfire, etc.
My last video upgrade was to an RX480, to get rid of all the heat from the 7970x2 crossfire setup. :)

But the death of Gamespy pretty much killed the year to year improvements for me, as all the games moved to "Developed for Xbox" instead of real gaming, like Crysis Wars.

Even the better games now are just Xbox handmedowns.

I had a lot of hope for DOOM, but between the levelup BS and the cheaters, there's no fun playing; we still stick with playing old games on LAN or VPN with closed servers.
Doom arcade mode, on Ultra-violence, is a lot of fun. :)

The most gaming fun I've had in years was the new Sigil levels released for Ultimate Doom; until we get away from the grinding games written by art institute rejects, there's no driver for new hardware for me.

If they want me to upgrade, put out a new, decent FPS with great graphics, a real anticheating system, developed for top end hardware, and M$ will have to release a real OS.
I don't see any of that happening, lol.
 
D

Deleted member 185158

Guest
Performance hasn't doubled in over 10 years; the performance increment has been ~15-20% between generations, if that.

On the topic, it's not about "performance", it's about density and transistor count mainly. We double the transistor count roughly every two years. (In reality a little bit longer on average)
 
Joined
Apr 21, 2008
Messages
5,250 (0.90/day)
Location
IRAQ-Baghdad
System Name MASTER
Processor Core i7 3930k run at 4.4ghz
Motherboard Asus Rampage IV extreme
Cooling Corsair H100i
Memory 4x4G kingston hyperx beast 2400mhz
Video Card(s) 2X EVGA GTX680
Storage 2X Crusial M4 256g raid0, 1TbWD g, 2x500 WD B
Display(s) Samsung 27' 1080P LED 3D monitior 2ms
Case CoolerMaster Chosmos II
Audio Device(s) Creative sound blaster X-FI Titanum champion,Creative speakers 7.1 T7900
Power Supply Corsair 1200i, Logitch G500 Mouse, headset Corsair vengeance 1500
Software Win7 64bit Ultimate
Benchmark Scores 3d mark 2011: testing
I have made a public speech in IT event exactly the same in the past April
 
Joined
Jul 16, 2014
Messages
8,115 (2.29/day)
Location
SE Michigan
System Name Dumbass
Processor AMD Ryzen 7800X3D
Motherboard ASUS TUF gaming B650
Cooling Artic Liquid Freezer 2 - 420mm
Memory G.Skill Sniper 32gb DDR5 6000
Video Card(s) GreenTeam 4070 ti super 16gb
Storage Samsung EVO 500gb & 1Tb, 2tb HDD, 500gb WD Black
Display(s) 1x Nixeus NX_EDG27, 2x Dell S2440L (16:9)
Case Phanteks Enthoo Primo w/8 140mm SP Fans
Audio Device(s) onboard (realtek?) - SPKRS:Logitech Z623 200w 2.1
Power Supply Corsair HX1000i
Mouse Steeseries Esports Wireless
Keyboard Corsair K100
Software windows 10 H
Benchmark Scores https://i.imgur.com/aoz3vWY.jpg?2
the appearance of chiplets, making system design seem more like a LEGO building,

I smell a new LEGO movie coming....

Moores law and transistor density vs. chip complexity and programming/compiler/microcode inefficiencies. I started out writing 6502 machine code. Lean. No wasted code, machine cycles, or memory. In 1980 those were scarce resources. We can develop code 100x faster today with modern languages and compilers, but they are in no way efficient compared to hard core machine code. I wonder, on average, exactly how inefficient modern compiled code is running in a typical OS, compared to effecient machine code. (Lost speed, wasted watts).
I had a C128, I wonder how much difference there was between the 6502 coding and coding on the CP/M side.
 
Last edited:
Top